[BUG] CUDA_TRY macro conflicts with other RAPIDS libraries

[jlowe]

Describe the bug
While trying to integrate RAPIDS RAFT and cuDF into the same application we ran across a compilation error due to a macro name conflict, e.g.:

[INFO]      [exec] In file included from /code/raft/cpp/include/raft/linalg/eig.cuh:20,
[INFO]      [exec]                  from /code/spark-rapids-ml/native/src/ml_utils.cu:6:
[INFO]      [exec] /code/raft/cpp/include/raft/cudart_utils.h:56: error: "CUDA_TRY" redefined [-Werror]
[INFO]      [exec]    56 | #define CUDA_TRY(call)                                                        \
[INFO]      [exec]       |
[INFO]      [exec] In file included from /miniconda3/include/cudf/utilities/type_dispatcher.hpp:22,
[INFO]      [exec]                  from /miniconda3/include/cudf/utilities/traits.hpp:20,
[INFO]      [exec]                  from /miniconda3/include/cudf/column/column_view.hpp:19,
[INFO]      [exec]                  from /miniconda3/include/cudf/column/column.hpp:18,
[INFO]      [exec]                  from /miniconda3/include/cudf/column/column_factories.hpp:18,
[INFO]      [exec]                  from /code/spark-rapids-ml/native/src/ml_utils.cu:1:
[INFO]      [exec] /miniconda3/include/cudf/utilities/error.hpp:102: note: this is the location of the previous definition
[INFO]      [exec]   102 | #define CUDA_TRY(call)                                            \
[INFO]      [exec]       |

Steps/Code to reproduce bug
Try to include raft/cudart_utils.h and cudf/utilities/error.hpp in the same compilation unit

Expected behavior
cudf macros do not conflict with definitions from other RAPIDS libraries (ideally with any library). Changing CUDA_TRY to CUDF_CUDA_TRY is one potential solution.

Additional context
rapidsai/raft#128 is the corresponding bug for RAFT

[github-actions]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[jlowe]

Still relevant and would like to see this fixed.

[jrhemstad]

Similar to #9797, we should rename to CUDF_CUDA_TRY.