[FEATURE]: Per-model cache efficiency and configuration diagnostics in Settings

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Settings > Models currently shows model names and visibility toggles, but it does not help users understand cache efficiency or diagnose the configuration behind each model. Cache data is available elsewhere as raw token counts or through CLI flags, while custom provider details may require editing opencode.json manually.

Add a per-model diagnostics and configuration view in Settings > Models that shows:

• cached input tokens and cache hit rate for a selectable time range
• input/output token usage and estimated cost
• provider, endpoint/protocol, and configuration source
• actions to edit the model/provider configuration and test the connection
• a clear "not reported by provider" state when cache metrics are unavailable

This would make it easier to compare models, verify that prompt caching is working, and fix provider configuration without switching between the UI, CLI statistics, and configuration files.

Related requests cover individual parts of this workflow: #42295 (session cache dashboard), #47481 (CLI cache statistics), #49319 (local model setup wizard), and #33704 (custom provider editing). This request is specifically for a unified per-model diagnostics and configuration surface in Settings.