Clean up Thrust includes. (rapidsai#4675)

## Description

This PR cleans up some `#include`s for Thrust. This is meant to help ease the transition to Thrust 1.17 when that is updated in rapids-cmake.

## Context

I opened a PR rapidsai/cudf#10489 that updates cuDF to Thrust 1.16. Notably, Thrust reduced the number of internal header inclusions:
> [rapidsai#1572](NVIDIA/thrust#1572) Removed several unnecessary header includes. Downstream projects may need to update their includes if they were relying on this behavior.

I spoke with @robertmaynard and he recommended making similar changes to clean up includes ("include what we use," in essence) to make sure we have compatibility with future versions of Thrust across all RAPIDS libraries.

This changeset also removes dependence on `thrust/detail` headers.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - William Hicks (https://github.com/wphicks)

URL: rapidsai#4675

cpp/src/arima/batched_arima.cu

@@ -20,6 +20,7 @@
 #include <vector>
 
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
 #include <thrust/fill.h>
 #include <thrust/for_each.h>
 #include <thrust/iterator/counting_iterator.h>

cpp/src/arima/batched_kalman.cu

@@ -20,6 +20,7 @@
 #include <cuml/tsa/batched_kalman.hpp>
 
 #include <cub/cub.cuh>
+#include <thrust/execution_policy.h>
 #include <thrust/for_each.h>
 #include <thrust/iterator/counting_iterator.h>
 

cpp/src/dbscan/corepoints/compute.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2022, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -18,6 +18,9 @@
 
 #include <raft/handle.hpp>
 
+#include <thrust/for_each.h>
+#include <thrust/iterator/counting_iterator.h>
+
 namespace ML {
 namespace Dbscan {
 namespace CorePoints {

cpp/src/dbscan/runner.cuh

@@ -34,6 +34,11 @@
 
 #include <label/classlabels.cuh>
 
+#include <thrust/copy.h>
+#include <thrust/device_ptr.h>
+#include <thrust/fill.h>
+#include <thrust/iterator/counting_iterator.h>
+
 #include <cstddef>
 
 namespace ML {

cpp/src/decisiontree/batched-levelalgo/kernels/quantiles.cu

@@ -16,6 +16,9 @@
  */
 #include "../quantiles.cuh"
 
+#include <thrust/execution_policy.h>
+#include <thrust/unique.h>
+
 namespace ML {
 namespace DT {
 

cpp/src/explainer/tree_shap.cu

@@ -28,6 +28,11 @@
 #include <raft/span.hpp>
 #include <rmm/device_uvector.hpp>
 #include <thrust/device_ptr.h>
+#include <thrust/device_vector.h>
+#include <thrust/execution_policy.h>
+#include <thrust/for_each.h>
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/memory.h>
 #include <treelite/tree.h>
 #include <type_traits>
 #include <variant>

cpp/src/fil/fil.cu

@@ -581,7 +581,7 @@ void check_params(const forest_params_t* params, bool dense)
   ASSERT(params->blocks_per_sm >= 0, "blocks_per_sm must be nonnegative");
   ASSERT(params->n_items >= 0, "n_items must be non-negative");
   ASSERT(params->threads_per_tree > 0, "threads_per_tree must be positive");
-  ASSERT(thrust::detail::is_power_of_2(params->threads_per_tree),
+  ASSERT((params->threads_per_tree & (params->threads_per_tree - 1)) == 0,
          "threads_per_tree must be a power of 2");
   ASSERT(params->threads_per_tree <= FIL_TPB,
          "threads_per_tree must not "

cpp/src/genetic/fitness.cuh

@@ -34,6 +34,8 @@
 #include <thrust/device_vector.h>
 #include <thrust/functional.h>
 #include <thrust/iterator/permutation_iterator.h>
+#include <thrust/memory.h>
+#include <thrust/scan.h>
 #include <thrust/sequence.h>
 #include <thrust/sort.h>
 #include <thrust/transform.h>

cpp/src/glm/qn/glm_base.cuh

@@ -24,6 +24,11 @@
 #include <raft/linalg/map_then_reduce.hpp>
 #include <raft/linalg/matrix_vector_op.hpp>
 #include <raft/stats/mean.hpp>
+
+#include <thrust/execution_policy.h>
+#include <thrust/functional.h>
+#include <thrust/reduce.h>
+
 #include <vector>
 
 namespace ML {

cpp/src/hdbscan/condensed_hierarchy.cu

@@ -27,9 +27,17 @@
 #include <raft/sparse/convert/csr.hpp>
 #include <raft/sparse/op/sort.hpp>
 
+#include <thrust/copy.h>
+#include <thrust/device_ptr.h>
 #include <thrust/execution_policy.h>
+#include <thrust/extrema.h>
+#include <thrust/functional.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/reduce.h>
+#include <thrust/sort.h>
+#include <thrust/transform.h>
 #include <thrust/transform_reduce.h>
+#include <thrust/tuple.h>
 
 #include <cuml/cluster/hdbscan.hpp>
 

cpp/src/hdbscan/detail/condense.cuh

@@ -30,7 +30,11 @@
 
 #include <cuml/cluster/hdbscan.hpp>
 
+#include <thrust/copy.h>
+#include <thrust/device_ptr.h>
 #include <thrust/execution_policy.h>
+#include <thrust/extrema.h>
+#include <thrust/fill.h>
 #include <thrust/reduce.h>
 
 namespace ML {

cpp/src/hdbscan/detail/extract.cuh

@@ -32,7 +32,9 @@
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>
 
+#include <thrust/device_ptr.h>
 #include <thrust/execution_policy.h>
+#include <thrust/extrema.h>
 #include <thrust/for_each.h>
 #include <thrust/reduce.h>
 #include <thrust/sort.h>

cpp/src/hdbscan/detail/membership.cuh

@@ -33,7 +33,9 @@
 #include <algorithm>
 
 #include <thrust/execution_policy.h>
+#include <thrust/fill.h>
 #include <thrust/for_each.h>
+#include <thrust/iterator/counting_iterator.h>
 #include <thrust/reduce.h>
 #include <thrust/sort.h>
 #include <thrust/transform.h>

cpp/src/hdbscan/detail/reachability.cuh

@@ -34,7 +34,10 @@
 #include <cuml/neighbors/knn.hpp>
 #include <raft/distance/distance.hpp>
 
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/transform.h>
+#include <thrust/tuple.h>
 
 namespace ML {
 namespace HDBSCAN {

cpp/src/hdbscan/detail/select.cuh

@@ -32,11 +32,18 @@
 
 #include <algorithm>
 
+#include <thrust/copy.h>
 #include <thrust/execution_policy.h>
+#include <thrust/fill.h>
 #include <thrust/for_each.h>
+#include <thrust/functional.h>
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/reduce.h>
 #include <thrust/sort.h>
 #include <thrust/transform.h>
+#include <thrust/transform_reduce.h>
+#include <thrust/tuple.h>
 
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>

cpp/src/hdbscan/detail/stabilities.cuh

@@ -33,10 +33,14 @@
 #include <algorithm>
 
 #include <thrust/execution_policy.h>
+#include <thrust/fill.h>
 #include <thrust/for_each.h>
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/reduce.h>
 #include <thrust/sort.h>
 #include <thrust/transform.h>
+#include <thrust/tuple.h>
 
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>

cpp/src/hdbscan/detail/utils.h

@@ -31,11 +31,16 @@
 
 #include "../condensed_hierarchy.cu"
 
+#include <thrust/copy.h>
 #include <thrust/execution_policy.h>
 #include <thrust/for_each.h>
+#include <thrust/functional.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/reduce.h>
 #include <thrust/sort.h>
 #include <thrust/transform.h>
+#include <thrust/transform_reduce.h>
+#include <thrust/tuple.h>
 
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>
@@ -186,4 +191,4 @@ void parent_csr(const raft::handle_t& handle,
 };  // namespace Utils
 };  // namespace detail
 };  // namespace HDBSCAN
-};  // namespace ML
+};  // namespace ML

cpp/src/hdbscan/runner.h

@@ -32,6 +32,10 @@
 #include "detail/reachability.cuh"
 #include <cuml/cluster/hdbscan.hpp>
 
+#include <thrust/device_ptr.h>
+#include <thrust/extrema.h>
+#include <thrust/transform.h>
+
 namespace ML {
 namespace HDBSCAN {
 

cpp/src/hierarchy/pw_dist_graph.cuh

@@ -33,8 +33,13 @@
 
 #include <thrust/device_ptr.h>
 #include <thrust/execution_policy.h>
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/iterator/zip_iterator.h>
 #include <thrust/scan.h>
+#include <thrust/sequence.h>
 #include <thrust/sort.h>
+#include <thrust/transform.h>
+#include <thrust/tuple.h>
 
 #include <limits>
 

cpp/src/kmeans/kmeans_mg_impl.cuh

@@ -20,6 +20,12 @@
 #include <rmm/device_scalar.hpp>
 #include <rmm/device_uvector.hpp>
 
+#include <thrust/execution_policy.h>
+#include <thrust/fill.h>
+#include <thrust/reduce.h>
+#include <thrust/scan.h>
+#include <thrust/transform.h>
+
 #include "common.cuh"
 #include "sg_impl.cuh"
 

cpp/src/kmeans/sg_impl.cuh

@@ -22,6 +22,9 @@
 #include <rmm/device_scalar.hpp>
 #include <rmm/device_uvector.hpp>
 
+#include <thrust/fill.h>
+#include <thrust/transform.h>
+
 namespace ML {
 
 namespace kmeans {

cpp/src/pca/sign_flip_mg.cu

@@ -27,6 +27,8 @@
 
 #include <thrust/device_vector.h>
 #include <thrust/execution_policy.h>
+#include <thrust/for_each.h>
+#include <thrust/iterator/counting_iterator.h>
 
 #include <cstddef>
 

cpp/src/randomforest/randomforest.cuh

@@ -16,17 +16,19 @@
 
 #pragma once
 
-#include <raft/common/nvtx.hpp>
-
 #include <decisiontree/batched-levelalgo/quantiles.cuh>
 #include <decisiontree/decisiontree.cuh>
 #include <decisiontree/treelite_util.h>
 
 #include <metrics/scores.cuh>
+
+#include <raft/common/nvtx.hpp>
+#include <raft/cudart_utils.h>
 #include <raft/random/permute.hpp>
 #include <raft/random/rng.cuh>
 
-#include <raft/cudart_utils.h>
+#include <thrust/execution_policy.h>
+#include <thrust/sequence.h>
 
 #ifdef _OPENMP
 #include <omp.h>

cpp/src/solver/lars_impl.cuh

@@ -37,6 +37,9 @@
 #include <rmm/device_uvector.hpp>
 #include <thrust/copy.h>
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
+#include <thrust/extrema.h>
+#include <thrust/fill.h>
 #include <thrust/iterator/counting_iterator.h>
 #include <thrust/sequence.h>
 #include <thrust/sort.h>

cpp/src/svm/linear.cu

@@ -36,8 +36,10 @@
 #include <rmm/device_uvector.hpp>
 #include <thrust/copy.h>
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
 #include <thrust/fill.h>
 #include <thrust/iterator/counting_iterator.h>
+#include <thrust/tuple.h>
 
 #include <glm/ols.cuh>
 #include <glm/qn/qn.cuh>

cpp/src/svm/smosolver.cuh

@@ -32,8 +32,11 @@
 #include <raft/cuda_utils.cuh>
 #include <raft/cudart_utils.h>
 #include <string>
+#include <thrust/copy.h>
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
 #include <thrust/fill.h>
+#include <thrust/iterator/constant_iterator.h>
 #include <type_traits>
 
 #include "kernelcache.cuh"

cpp/src/svm/svc_impl.cuh

@@ -38,6 +38,7 @@
 #include <rmm/mr/device/per_device_resource.hpp>
 #include <thrust/copy.h>
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
 #include <thrust/iterator/counting_iterator.h>
 
 namespace ML {

cpp/src/svm/workingset.cuh

@@ -36,7 +36,9 @@
 #include <thrust/device_ptr.h>
 #include <thrust/iterator/permutation_iterator.h>
 
+#include <thrust/copy.h>
 #include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
 #include <thrust/iterator/permutation_iterator.h>
 
 #include <cub/cub.cuh>

cpp/src/tsne/barnes_hut_tsne.cuh

@@ -22,6 +22,10 @@
 #include <raft/cudart_utils.h>
 #include <raft/linalg/eltwise.hpp>
 
+#include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
+#include <thrust/fill.h>
+
 namespace ML {
 namespace TSNE {
 

cpp/src/tsne/distances.cuh

@@ -34,6 +34,9 @@
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>
 
+#include <thrust/functional.h>
+#include <thrust/transform_reduce.h>
+
 namespace ML {
 namespace TSNE {
 

cpp/src/tsne/exact_kernels.cuh

@@ -21,6 +21,10 @@
 #include <raft/cudart_utils.h>
 #include <raft/linalg/eltwise.hpp>
 
+#include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
+#include <thrust/reduce.h>
+
 #define restrict __restrict__
 
 namespace ML {

cpp/src/tsne/exact_tsne.cuh

@@ -20,6 +20,10 @@
 #include <cuml/common/logger.hpp>
 #include <raft/cudart_utils.h>
 
+#include <thrust/device_ptr.h>
+#include <thrust/execution_policy.h>
+#include <thrust/fill.h>
+
 namespace ML {
 namespace TSNE {