fix(grokbuild): bind session summaries to custom provider profiles

[enderzcx]

Summary / 概述

Grok Build chooses the model for session titles/summaries independently of the main session (models.session_summary vs models.default). Generated custom-provider configs currently omit the former, so a working custom-model conversation can still issue a title request to the client's fallback model.

This focused fix uses the existing configuration builders:

• Set models.session_summary to the selected profile when generating Grok Build deep-link and form configurations.
• Preserve a separately configured summary profile when editing a provider.
• If summaries followed the previous default profile, follow its rename instead of referring to the model table that the editor removes.
• Use the profile name, not the upstream model ID, so aliases resolve through the correct provider table.

No new UI, dependency, model catalog changes, credential-resolution changes, API retries, or provider protocol changes. Existing deep-link env_key protections are unchanged and their tests still pass.

Reference: Grok Build model configuration.

Related Issue / 关联 Issue

Fixes #7003

Verification / 验证

Run locally on macOS:

• pnpm typecheck — PASS.
• pnpm format:check — PASS.
• pnpm exec vitest run src/utils/grokBuildConfig.test.ts tests/components/GrokBuildProviderForm.test.tsx tests/utils/deepLinkConfigPreview.test.ts --reporter=dot — 21 PASS.
• cd src-tauri && cargo fmt --check — PASS.
• cargo test --locked --lib deeplink::provider::tests::grokbuild --no-fail-fast — 3 PASS, including both existing credential-import regression tests. The new test covers selected, trimmed, namespaced, absent and blank-model behavior through the configuration builder.
• cargo clippy --locked --all-targets --all-features — PASS.
• git diff --check — PASS.

The new assertions were first run against the original builders: four TypeScript checks and the Rust summary-field check failed because the field was missing. They pass with this patch.

Full frontend suite: pnpm test:unit --maxWorkers=2 --minWorkers=1 ran 1,016 tests: 1,014 passed, 2 failed in the unchanged tests/config/codexReasoningLevelPresets.test.ts (xAI (Grok) and xAI (Grok) OAuth, grok-4.5: actual reasoning levels include xhigh, expected levels do not). Both failures were reproduced in a separate unmodified 054673e0b8cca7d8bd8c67658f2c28e63b2855e0 worktree by running that file: 38 passed, the same 2 failed. No reasoning catalog or unrelated test assertions were changed.

This is a config-generation regression fix, not a claim of complete desktop GUI or cross-platform acceptance. The patch was AI-assisted; the code and changes were reviewed and the checks above were actually run locally.

Screenshots / 截图

Not applicable: no visual changes.

Checklist / 检查清单

• pnpm typecheck passes / 通过 TypeScript 类型检查
• pnpm format:check passes / 通过代码格式检查
• cargo clippy passes (if Rust code changed) / 通过 Clippy 检查
• No user-facing text changed; i18n updates are not needed / 未修改用户可见文案，无需更新国际化

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-08-31T02:43:33.133238Z	f712134	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.