add mmproj vision support

experimental boosting changes

Use temp download dir and robust rmtree handler to avoid permission errors on cleanup

Save models using repo filename, set MODEL_PATH from marker, ensure HF xet log dirs writable

Allow MODEL_FILE override; improve diagnostics when no .gguf files present

Add QUANT_PREFERENCE and PERF_PROFILE (low_latency/balanced/throughput) with server defaults

Add model list and download progress logging

Speed up model download by selecting one GGUF

Add optimized llama.cpp CPU inference backend