feat: add openai_compatible LLM profile

[dimknaf]

Adds an openai_compatible profile so the internal agent can talk to any OpenAI-compatible /v1 endpoint — Ollama, LM Studio, copilot-api, a remote vLLM — without introducing another provider abstraction.

LLM_PROFILE=openai_compatible
AGENT_MODEL=openai/llama3.2:3b
OPENAI_BASE_URL=http://host.docker.internal:11434/v1
OPENAI_API_KEY=                 # only if the endpoint requires auth

Why this shape

• Reuses the existing mechanism. It's one entry in _LLM_PROFILES (model-prefix + api_key_env) plus LiteLLM's openai/ routing the vllm_* profiles already use — no new code path.
• Env-driven base URL, scoped to this profile only. resolved_base_url special-cases openai_compatible; every other profile's base_url stays fixed in the table, so nothing can be silently re-pointed by an env var.
• Provider-prefix env names (OPENAI_BASE_URL / OPENAI_API_KEY) to match DEEPINFRA_API_KEY, NVIDIA_NIM_API_KEY, VLLM_API_KEY — instantly recognizable to anyone who's used an OpenAI-compatible tool.

Changes

braindb/config.py (profile + scoped resolved_base_url + EMPTY-key fallback for the env endpoint), docker-compose.yml (env passthrough), .env.example, README.md, CONTRIBUTING.md, CHANGELOG.md, and tests/test_config_profiles.py.

Verification

• New unit tests pass stack-free, including a regression that OPENAI_BASE_URL does not leak into other profiles.
• Full suite green against the isolated test stack; docker compose config parses with the new env present.

Reworked from @WarGloom's PR #7 — thanks for the nudge and the original direction. Adjusted to the provider-prefix env convention and a profile-scoped override per the review there.