Adds caching to get_encoding to avoid repeatedly constructing Encodings

[tal7aouy]

Summary

This PR adds caching to get_encoding to avoid repeatedly constructing Encodings.

Implementation

• Adds @lru_cache decorator to cache get_encoding return values
• Sets maxsize=None to cache all encodings
• First call constructs encoding as normal
• Subsequent calls hit cache and avoid reconstructing
• Locking and global state still handled properly

Benefits

• Significant speedups when requesting the same encodings multiple times

• Simple change with minimal code changes

• Fully backwards compatible

  Future Improvements

• Add max cache size to bound memory usage

• Support cache keying off config instead of just name

Overall this simple change provides large performance improvements by caching encoding objects.

Please let me know if any changes are needed!

[hauntsaninja]

There is already caching