feat(sse): release GPT-6 Sol and Luna support

[TonisOrmisson]

Summary

• Port GPT-6 Sol and Luna Codex support from feat(sse): add GPT-6 Sol and Luna to the Codex catalog diegosouzapw/OmniRoute#14677: catalog variants, effort aliases/clamping, Responses translation, pricing, Fast costs and VS Code metadata.
• Port the related reasoning-routing/editor fix from fix(routing): read Codex max/ultra models from the alias sets in reasoning rules diegosouzapw/OmniRoute#14720 so Max/Ultra rules accept the new models.
• Preserve existing GPT-5.6/Astra behavior and the fork's prebuilt deployment workflow.
• Repair the transitive brace-expansion lock mismatch that blocked npm ci in today's release build.
• Integrate the separate base corrections in fix(ci): clear inherited release validation failures #3 (expired test data and two inactive foreign-fork publishers). No unrelated production behavior changes.
• GPT-6 Terra is not added: no confirmed catalog entry was found in the inspected upstream source/PR.

Upstream implementation and model regression tests by Quang Blue, credited in the commit.

Validation

• RED/GREEN: 10 new catalog tests and 8 new reasoning-routing tests reproduced the missing behavior before the port and pass afterward.
• Focused GPT-5.6/Astra/Codex/routing run: 40 passed.
• Vitest: 44 files, 405 passed (rerun on the integrated candidate).
• Full native unit phase: 34,387 tests; 34,364 passed, 18 skipped, 5 failed on that run. Two failures were the inherited foreign-workflow policy checks, now fixed separately in fix(ci): clear inherited release validation failures #3. The three timing/cache tests pass in an isolated rerun without parallel-suite load.
• Remaining native unit phases ran separately because the first phase's nonzero result stops npm's && chain: dashboard 125 passed; serial 23 passed.
• Final validation on the integrated release head: all 39 new/previously failing tests passed. No unresolved failure among the tested files; the monolithic unit command is NOT being represented as an all-green single run.
• npm run typecheck:core: passed.
• Changed-file ESLint with repository suppressions, Prettier, pre-commit gates: passed. All hooks executed; lint-staged used --no-stash for shared-worktree safety.
• Provider consistency check: passed via Node (the installed Bun binary is unavailable locally).
• npm ci --dry-run --ignore-scripts --no-audit --no-fund: failed before the lock correction and passes afterward.
• Production Build App and Build Prebuilt Standalone succeeded for the production-code change (751afb6). Prebuilt also succeeded for e3577c4; later changes only remove inactive workflow files.
• Downloaded prebuilt: BUILD_SHA matches the requested commit; bundled server chunks contain Sol/Luna and their Max/Ultra variants. Local isolated startup returned HTTP 200 from /api/health. No live Codex credential call or production deployment was performed.

Release

Use the existing to/main push workflow to build and publish from the exact final commit. Do not manually publish a tag using the old committed prebuilt chunks.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.