v1.6.0 - Multi-Provider LLM Support

Multi-Provider LLM Support

Hermes Life OS no longer requires an OpenRouter API key. It now runs on
four backends, auto-detected from whichever key you have set:

• 🦙 Ollama - free, fully local, no API key needed
• 🟣 Anthropic
• 🟢 OpenAI
• 🔀 OpenRouter (original default, including Hermes-3)

What's new

• New demo/llm_providers.py: a provider-agnostic client layer with
  auto-detection (Anthropic > OpenAI > OpenRouter > Ollama) and an
  Anthropic ↔ OpenAI message/tool format adapter, so the rest of the
  codebase didn't need to change.
• --provider flag / LIFE_OS_PROVIDER env var to force a specific
  backend regardless of which keys are set.
• Friendly troubleshooting output on connection/auth failures instead
  of a raw traceback (e.g. "ollama isn't running — try ollama serve").
• Updated README.md and docs/SETUP.md with setup instructions for
  all four backends.

Testing

16 new unit tests covering provider resolution and the Anthropic format
adapter (fully mocked, no network calls). Full suite: 129/129 passing
(113 pre-existing + 16 new).