[BUG] skip_rows doesn't work properly in ChunkedParquetReader

[lithomas1]

Describe the bug
A clear and concise description of what the bug is.

I'm getting bad CUDA behavior like

RuntimeError: parallel_for failed: cudaErrorInvalidDevice: invalid device ordinal

when skiprows is greater than 0
nrows looks OK, though.

I don't think this is an issue with my GPU since other tests like the pylibcudf I/O tests pass.

Steps/Code to reproduce bug
Follow this guide http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports to craft a minimal bug report. This helps us reproduce the issue you're having and resolve the issue more quickly.

Checkout my branch https://github.com/lithomas1/cudf/tree/parquet-nrows-bug
which adds bindings for the chunked parquet reader nrows/skip_rows options

and then run the test

pytest python/cudf/cudf/tests/test_parquet.py -v -k "test_parquet_chunked_reader_nrows_skiprows"

Expected behavior
A clear and concise description of what you expected to happen.

The included test should pass.

Environment overview (please complete the following information)

• Environment location: [Bare-metal, Docker, Cloud(specify cloud provider)]
• Method of cuDF install: [conda, Docker, or from source]
  • If method of install is [Docker], provide docker pull & docker run commands used

Environment details
Please run and paste the output of the cudf/print_env.sh script here, to gather any other relevant environment details

Additional context
Add any other context about the problem here.

I think this is also related to the chunk_read_limit/pass_read_limit.
When those are set to 0, the test passes.
(but maybe that is because that goes through a separate path, I haven't looked at the C++ code for this)

[lithomas1]

duplicate of #16186