vLLM Benchmarking

Provides interactive benchmarking for vLLM through the Model Context Protocol. Allows users to specify endpoints, models, iteration counts, and prompt numbers for performance testing.