Test all keys missed in storage in caching and eval mode

[jiashuy]

Description

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[greptile-apps]

Greptile Summary

This PR adds a defensive check to prevent calling cache operations with empty tensors. In eval mode, when all keys are missing from storage, the code would previously attempt to update the cache with empty key tensors, which could cause issues. The fix adds an early return in update_cache() when there are no keys to process.

• Added early return check in update_cache() when missing_keys.numel() == 0
• Added test coverage for the corner case where all requested keys are missing in eval mode
• The fix prevents potential issues with cache.insert_and_evict() being called with empty tensors

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• The change is a simple defensive check that prevents edge case errors, includes appropriate test coverage, and follows good engineering practices. The logic is straightforward and the test validates the fix works correctly.
• No files require special attention

Important Files Changed

Filename	Overview
corelib/dynamicemb/dynamicemb/key_value_table.py	Added early return in update_cache() to handle empty missing_keys tensor, preventing potential issues with cache operations
corelib/dynamicemb/test/test_batched_dynamic_embedding_tables_v2.py	Added test case for corner case where all keys are missing from storage in eval mode

_{Last reviewed commit: 5f44bc6}

[greptile-apps]

_{2 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[jiashuy]

There is no problem even if I remove the check. Adding it to get more robust.

[shijieliu]

merge this PR into #313 and close this one.

[jiashuy]

closed as already fixed in #349