[Feat] BYOK: Add Gateway provider for custom inference endpoints

[abgunes]

Is your feature request related to a problem? Please describe.
The Model Management preference page lists a fixed set of providers (Azure, OpenAI, Anthropic, OpenRouter, Gemini, Groq). There is no way to add a model from a custom inference endpoint, such as an enterprise API gateway or a locally running inference server, because every provider targets a hard-coded upstream host.

Describe the solution you'd like
Add a Gateway entry to the provider list alongside the existing ones. When the user selects Gateway and clicks "Add Model...", a credentials dialog should collect:

• Gateway base URL (e.g. https://my-gateway.internal/v1 or http://localhost:11434)
• API key with show/hide toggle (optional for local servers)
• Auth scheme selector: x-api-key header or Authorization: Bearer
• Model discovery toggle to fetch available models from {baseUrl}/v1/models automatically

The rest of the flow (Remove Model, Enable, Reload, Change API, Delete API) stays exactly as it is for all other providers.

Describe alternatives you've considered
There is no working alternative within the current UI. The existing providers cannot be redirected to a custom endpoint. Configuring an HTTP proxy is the only partial workaround but it adds infrastructure complexity and does not cover local servers on loopback.

Additional context
The screenshot below shows the current provider list. The request is to add Gateway as a seventh entry in the same list, with no changes to the existing provider flows.