refactor(http_parsing_utils.py): streamline request body handling and…

… memory management

- Removed the weak reference cache for request bodies and replaced it with direct storage on the request object to prevent memory leaks.
- Implemented immediate garbage collection every 100 requests to manage memory usage effectively.
- Added a cleanup function to explicitly free memory associated with request processing.
- Updated the logic for retrieving and storing parsed request bodies to enhance performance and reliability.

Add caching for failed deployments due to configuration errors

- Introduced a new cache to track deployments that fail due to configuration issues, preventing repeated retry attempts.
- Implemented logic to skip deployments that have previously failed due to configuration errors.
- Added a method to clear the failed deployments cache, allowing for retries of specific or all failed deployments.

Implement pass-through route registration and removal to prevent dupl…

…icates and memory leaks

fix memory leak issue

fixes for update_group

fix

fix

fix(health_check_helpers.py): set max tokens for wildcard call to 10,…

… fixes calling gpt-5-nano via wildcard on openai (BerriAI#13482)

gpt-5-nano raises errors for max_tokens=1

bump: version 1.75.4 → 1.75.5

added mcp guardrails doc in mcp.md (BerriAI#13452)