feat: add LiteLLM unified LLM provider (RFC-012) (#108)

Introduces a new provider: litellm that routes to 100+ LLM backends
through LiteLLM's model-name prefix matching. Single provider replaces
the need for separate openai_compat, anthropic, bedrock, vertex, and
similar dedicated providers.

Changes:
- Create src/zettelforge/llm_providers/litellm_provider.py with
  LiteLLMProvider class implementing the LLMProvider protocol
- Register 'litellm' provider in __init__.py (conditional on SDK)
- Add litellm>=1.60.0 optional dependency to pyproject.toml
- Update config.default.yaml with provider docs and examples
- Add 11 unit tests: construction, generate, api_key, system prompt,
  json_mode, protocol satisfaction, registration check
- Remove placeholder lines for openai_compat and anthropic (now
  superseded by LiteLLM)
- Create docs/rfcs/RFC-012-litellm-unified-provider.md (369 lines)

RFC: docs/rfcs/RFC-012-litellm-unified-provider.md

Author: rolandpg

config.default.yaml

@@ -148,9 +148,8 @@ embedding:
 # Provider selects the backend registered in zettelforge.llm_providers:
 #   local           — in-process LLM inference (see local_backend below; requires zettelforge[local] or zettelforge[local-onnx])
 #   ollama          — Ollama HTTP API
+#   litellm         — LiteLLM routing to 100+ providers (RFC-012; requires zettelforge[litellm])
 #   mock            — test-only canned responses
-#   openai_compat   — any /v1/chat/completions endpoint (Phase 2, upcoming)
-#   anthropic       — Anthropic native SDK (Phase 3, upcoming)
 #
 # local_backend selects the in-process inference engine when provider=local:
 #   llama-cpp-python     — GGUF models via llama-cpp-python (default; requires zettelforge[local])
@@ -174,6 +173,19 @@ embedding:
 #     filename: phi3-mini-4k-instruct-q4.onnx
 #     provider: rocm            # cpu, cuda, dml, rocm, openvino, coreml
 #
+#   # LiteLLM — unified routing (requires pip install zettelforge[litellm])
+#   # Model name determines the provider. Examples:
+#   #   gpt-4o, gpt-4o-mini                      → OpenAI
+#   #   claude-sonnet-4-20250514, claude-3-opus   → Anthropic
+#   #   gemini/gemini-2.0-flash                   → Google Gemini
+#   #   groq/llama-3.3-70b-versatile              → Groq
+#   #   together_ai/meta-llama/Llama-3.3-70B      → Together AI
+#   #   openrouter/anthropic/claude-3.5-sonnet    → OpenRouter
+#   #   bedrock/anthropic.claude-3-sonnet         → AWS Bedrock
+#   provider: litellm
+#   model: gpt-4o
+#   api_key: ${OPENAI_API_KEY}
+#
 # api_key supports ${ENV_VAR} references — never commit raw API keys.
 #   api_key: ${OPENAI_API_KEY}
 #