Merge pull request #2 from garrettyarmo/feature/elevenlabs-provider

rename vapi -> provider in pipeline.py for provider-agnostic clarity

remove redundant 'is not None' check: prev_time is always a float

remove unused end_phrases: ElevenLabs handles conversation termination server-side

add configurable simulate_timeout_secs for ElevenLabs conversations (default 300s)

revert load_dotenv override=True: preserve standard env var precedence

update README: add ElevenLabs to providers, setup, and project structure

fix ElevenLabs provider: response parsing, dynamic variables, timeout

add ElevenLabs ConvAI provider

gitignore all results_* directories

redesign score progression graph: Karpathy style with gray discards, green keeps, step-line running best

add autoresearch graphs (score progression, metrics panel, keep/discard, prompt evolution, failure modes), speed up Vapi conversations

remove stale results and launch post from repo

add 'python main.py results' command: post-run summary with score progression, kept changes, best prompt, failure modes

rewrite examples and README: salon (Vapi) + pizza delivery (Smallest AI), explain description-driven attack generation