fix(frontend): persist model selection per thread

[LittleChenLiya]

Summary

• persist model selection per thread in local storage
• load thread-specific model settings when switching chats or agents

Fixes #1541

[Copilot]

Pull request overview

Persists the selected model on a per-thread basis in the frontend so switching between chat threads/agents restores each thread’s own model choice instead of sharing a single global selection.

Changes:

• Add per-thread localStorage keys for model_name and load them when a threadId is provided.
• Update useLocalSettings to accept an optional threadId and persist per-thread model changes.
• Wire thread-aware settings into both the generic chat page and agent chat page.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
frontend/src/core/settings/local.ts	Introduces per-thread model storage and extends getLocalSettings/saveLocalSettings to optionally scope model selection by thread.
frontend/src/core/settings/hooks.ts	Updates useLocalSettings to accept threadId and persist/load settings accordingly.
frontend/src/app/workspace/chats/[thread_id]/page.tsx	Uses thread-aware local settings in the main chat page.
frontend/src/app/workspace/agents/[agent_name]/chats/[thread_id]/page.tsx	Uses thread-aware local settings in the agent chat page.

[foreleven]

I think this would be cleaner as a separate useThreadSettings hook rather than extending useLocalSettings with thread-specific behavior. useLocalSettings reads like global user settings, but this logic only applies when a thread is present and only affects that thread’s selection. Keeping them separate would make the API clearer and avoid mixing global settings with per-thread overrides.

[LittleChenLiya]

已按建议拆分，新增了 useThreadSettings 来承载 thread-specific 的 model override，useLocalSettings 现在只保留全局设置语义，这样职责会更清晰一些。