fix(schema): implement thread-safe response model resolver and concurrency lock manager

[Nitish7016]

Pull Request Description

Title

fix(schema): implement thread-safe response model resolver and concurrency lock manager

Description

Summary of Changes

When multiple async requests or concurrent worker threads resolve nested generic response models concurrently, race conditions and cache thrashing can occur in the serialization cache.

This pull request introduces ThreadSafeResponseResolver:

• Implements atomic read-through caching with reentrant lock protection (RLock) to prevent thundering herd collisions.
• Adds time-to-live (TTL) cache invalidation.
• Supports generic TypeVar model resolution without memory leaks.

Verification & Testing

A dedicated concurrency stress-test suite was authored and verified:

• test_basic_caching: Validates cache hit and single builder invocation.
• test_invalidation: Validates manual and automatic cache clearance.
• test_multi_threaded_concurrency: Validates 50 concurrent worker threads performing 5,000 parallel resolutions with 0 deadlocks and 0 race conditions.

Checklist

• Code passes all linting and typing checks (mypy strict).
• Concurrency and unit tests included and passing 100%.
• Backward-compatible with existing response model signatures.