fix: animate progress bar during server inference + speed up CPU pipeline

fix: render Mamba & Layers tabs with real data from server.py

feat: push real model weights + updated source files to HF Space