feat(ai): replace OpenWebUI provider with LiteLLM

[dave6892]

Summary

• Deprecates the OpenWebUI provider variant and replaces it with LiteLLM
• LiteLLM proxy is fully OpenAI-compatible, so chat and model-listing use the standard /v1/chat/completions and /v1/models endpoints
• Default base URL updated to http://localhost:4000 in the Settings UI
• Existing configs with provider: openwebui fall back to Ollama on next load; users re-save via Settings

Changes

File	Change
crates/types/src/config.rs	Openwebui → Litellm in AiProvider enum
crates/ai/src/stream.rs	chat_url and list_models updated; list_models_openwebui replaced with list_models_litellm
apps/desktop/src-tauri/src/commands.rs	save_ai_config / get_ai_config map "litellm" ↔ AiProvider::Litellm
apps/desktop/src/views/SettingsView.tsx	Provider type, defaults, button array, and label updated to LiteLLM

Test plan

• cargo build -p nichinichi-types -p nichinichi-ai -p nichinichi-cli passes
• Start LiteLLM proxy on localhost:4000, select LiteLLM in Settings, fetch models list
• Run nichinichi ask "test" with provider: litellm in ~/.nichinichi.yml
• Verify Ollama provider still works unchanged

🤖 Generated with Claude Code