improvement(mothership): relabel the Sol pick GPT-6.1 Sol and retire the none effort - #8752

Summary

• The Mothership worker now serves the composer's Sol pick with OpenAI's GPT-6.1 Sol, so the model picker labels it GPT-6.1 Sol. The wire value stays gpt-6-sol, and the shared protocol contract is unchanged.
• GPT-6.1 Sol has no none reasoning effort (it supports low through max), and no other composer model does either. The picker drops its None option and every model now shows the same effort list.
• resolveMothershipModelSettings now lifts none to low. The composer, the client send path and server admission all go through it, so a stored none (a chat's saved effort in copilot_chats.config, or a request from a stale client) runs and displays as Low. Admission records low as the chat's choice on that chat's next send. Before this change, none fell back to the high default when the model picker was off or when the pick wasn't Sol.
• Sim shows no Sol pricing or cost estimate, and no docs page mentions the picker, so there is nothing else to update.

Type of Change

• Improvement

Testing

• post.test.ts: the effort table now covers a Sol chat with a stored none and a Sol send of none. Both run at low, and low is recorded as the chat's choice. A submitted none with the picker off also runs at low. All three cases fail without the change.
• set-effort.integration.ts (real Postgres): a chat stored with none runs its next turn at low. It fails without the change.
• bun run lint, bun run type-check, bun run check:audits, docs-manifest:check, the block-registry check, and root bun run test all pass.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing (new tests pass the test-audit authoring gate)
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)