Add OpenAI and OpenRouter model profile support

## Goal\n\nSupport OpenAI and OpenRouter as selectable model-profile providers so operators can use their existing AI API accounts for supervisor evaluation, Cycle Improvement AI, and the built-in Chat Assistant.\n\n## Scope\n\n- Add provider settings and validation for OpenAI-compatible API credentials.\n- Add OpenAI and OpenRouter choices to the model-profile UI.\n- Keep credentials in environment variables; persist only the variable name and non-secret connection settings.\n- Add provider tests and update the BYOA documentation.\n\n## Acceptance criteria\n\n- An operator can configure an OpenAI profile with an API-key environment variable and model name.\n- An operator can configure an OpenRouter profile with an API-key environment variable, model name, and optional endpoint headers.\n- Both providers work anywhere a System AI model is selected.\n- Existing Azure OpenAI and Bedrock profiles remain compatible.

Start by locating the existing Azure OpenAI and Bedrock model-profile providers, their settings and validation, the model-profile UI, provider tests, and the BYOA documentation. Trace how a System AI model is selected, then verify both new profiles across the listed use cases while preserving existing provider compatibility.

Written by the indexing model from the issue text.