Add regex-based enum extraction to fix header parsing

[devin-ai-integration]

Description

closes NVI-3

This PR replaces the hardcoded enum handling in cuda_bindings/setup.py with dynamic regex pattern matching to find CUDA enum types that pyclibrary.CParser fails to extract.

Changes

• Added extract_enums_from_headers() function with three regex patterns to match:
  • typedef __device_builtin__ enum X Y; (CUDA-specific enum typedefs)
  • typedef enum X Y; (standard enum typedefs)
  • enum X {...} Y?; (inline enum declarations)
• Integrated regex extraction into parse_headers() after CParser processing
• Removed the manual CUstreamAtomicReductionDataType_enum addition and associated TODO comment
• Preserved existing hardcoded replacements in the replace dictionary for backward compatibility

Human Review Checklist

• Verify the regex patterns correctly match CUDA enum declarations in actual headers
• Confirm CUstreamAtomicReductionDataType_enum is now found by the regex extraction (previously required manual addition)
• Consider whether duplicate entries in found_types (from both CParser and regex) could cause downstream issues
• The third regex pattern uses [^}]* which won't handle nested braces - verify this is acceptable for CUDA enum definitions

Checklist

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

---

Link to Devin run
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