cuda.bindings: Fix API status handling

[juenglin]

closes#2299

Test get_bar_size_in_kb failed on an GH200 system. The clue is in the error shared in #2299:

tests/test_cufile.py::test_get_bar_size_in_kb FAILED
bar_size_kb = cufile.get_bar_size_in_kb(0)
cuda.bindings.cufile.cuFileError:
SUCCESS (0): cufile success; CUDA status: CUDA_ERROR_NOT_SUPPORTED (801)

The cufile status struct has two fields, one for cufile proper, and a second for forwarding potential CUDA driver errors. The CUDA driver error field should be ignored unless the cufile error is CU_FILE_CUDA_DRIVER_ERROR. So the test failed because cuda.bindings incorrectly raised a cuFileError.

The issue fix is here. The PR contains other unrelated changes from re-generating cufile bindings.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[mdboom]

LGTM.

The unrelated changes make sense given the recent generator changes. We should probably do a "catch-up" PR for all of the libraries, but that can be a separate PR.

[juenglin]

/ok to test

[copy-pr-bot]

/ok to test

@juenglin, there was an error processing your request: E1

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/1/

[juenglin]

/ok to test 9ef5766

[github-actions]

Doc Preview CI
Preview removed because the pull request was closed or merged.