AI Automation Suggester 1.5.3

This patch focuses on provider setup reliability and endpoint flexibility for local and OpenAI-compatible deployments.

Highlights

Added an optional Ollama/Open WebUI base URL field. You can still use the existing host, port, and HTTPS fields for native Ollama, or provide a full base URL such as http://homeassistant.local:3000/ollama.
Added fallback Ollama API path handling for native Ollama and Open WebUI proxy paths.
Made Custom OpenAI-compatible validation accept base URLs, /v1 URLs, exact chat completion URLs, and common model-listing endpoints.
Switched Anthropic and Gemini setup checks to model-listing endpoints where possible, avoiding fragile one-token generation pings during setup.
Applied the configured request timeout to more setup validation calls, including OpenRouter.
Added endpoint normalization regression tests.

Existing Ollama configurations continue to work with host, port, and HTTPS settings.
The new base URL field takes precedence when it is provided.
Larger feature requests are tracked in docs/FEATURE_REQUEST_PLAN.md.