Fix enum parsing for CUstreamAtomicReductionDataType_enum

[devin-ai-integration]

Description

closes NVIDIA#1312

This PR fixes the parse_headers function in cuda_bindings/setup.py to properly extract CUstreamAtomicReductionDataType_enum from CUDA headers, eliminating the manual workaround.

Root cause: pyclibrary's CParser fails to parse enums when their values use unsigned literal shift syntax (e.g., 1u<<0). In CUDA 13.x headers, CUatomicCapability_enum uses 1u<<0 syntax, and CUstreamAtomicReductionDataType_enum references those values with bitwise OR expressions, causing the parser to fail.

Fix:

1. Add a replace pattern "1u<<": "1<<" to normalize unsigned literal shifts before parsing
2. Add safety net to extract enum tag names from defs["types"] keys that start with "enum " prefix (pyclibrary sometimes stores enums there instead of defs["enums"])

Validation: Tested against actual CUDA 13.1.0 headers downloaded from NVIDIA's redistribution server, confirming the enum is now correctly extracted.

Checklist for human review

• Verify the 1u<< replacement doesn't affect other header parsing unexpectedly
• Confirm CI passes with CUDA 13.x headers

Checklist

• New or existing tests cover these changes.
• The documentation is up to date with these changes.

---

Link to Devin run: https://app.devin.ai/sessions/349f9751c64e4be08f02bb35be35eb52
Requested by: Shawn Azman (shawn@cognition.ai) (@ShawnAzman)

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring