Add concurrent benchmark script (benchmark.py) for heavy load testing with per-method performance metrics including success rate, latency percentiles, and requests-per-second. Introduce enhanced server variant (main3.py) implementing "Prime on Write" metadata cache strategy for consistently fast PROPFIND responses, alongside stability fixes. Include secondary benchmark variant (benchmark2.py) with filename tracking for created/moved resources, and a baseline server implementation (main2.py) with multi-layered caching via ETags, LRU metadata cache, and async file operations.
