feat(asr): add optional Nemotron voice-note input

perf(modal): bump per-server max_model_len to 8k/4k

perf(agent,modal): aggressive summarization + per-server max_model_len

perf(modal): bake weights, eager-load, drop max-model-len, AOT cache

perf(modal): cold-start on demand with eager-mode and warmup ping

Fabella Phase 3: chat UI, HF Auth, bucket JSON history

Fabella Phase 2: add read aloud and redesign frontend