docs(readme): add omlx (Apple Silicon MLX) to local runtime list

[fredchu]

Documentation-only. Lists omlx as an OpenAI-compatible local runtime alongside LM Studio, Ollama, and llama.cpp server, in all 8 README translations. README.md and README.zh-TW.md (the two files with "Runtime notes" sections) also get a runtime-specific bullet describing the startup shape.

What

• All 8 READMEs: add omlx (Apple Silicon) to the "local OpenAI-compatible runtimes" feature bullet.
• All 8 READMEs: add omlx: http://localhost:8000/v1/chat/completions to the endpoint preset code block.
• README.md + README.zh-TW.md: add one Runtime-notes bullet describing omlx serve / omlx serve --model-dir /path/to/models startup, with the OpenAI-compatible endpoint at http://localhost:8000/v1/chat/completions.

Why

omlx is an Apache-2.0 MLX inference server optimized for Apple Silicon. Its OpenAI-compatible API defaults to http://localhost:8000/v1/chat/completions, matching the existing preset format. Margin users on M-series Macs are a natural audience for it.

No code changes are required — the runtime is reached via the existing OpenAI Compatible provider with a user-entered endpoint.

Alignment

• PRINCIPLES.md → Keep Provider Choice Modular: documents another OpenAI-compatible local endpoint without adding a new provider adapter.
• PRINCIPLES.md → Avoid Backend Lock-In: still uses the user-configured endpoint path; no official backend introduced.
• PRINCIPLES.md → Keep the MVP Narrow: no scope expansion (no new feature, no new file type).

Translation discipline

• The 6 truncated translations (zh-CN, ja, ko, de, es, fr) — which don't currently contain a "Runtime notes" section — only get the features-bullet + code-block changes, matching their existing structure.
• The de/es/fr READMEs intentionally use ASCII-only spelling; the added text stays ASCII to match.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Documentation-only update that adds omlx (an Apple Silicon MLX inference server) as another OpenAI-compatible local runtime example across all 8 README translations. No code changes are required because the existing openai-compatible provider already accepts arbitrary user-configured endpoints.

Changes:

• Adds omlx (Apple Silicon) to the local-runtime features bullet in all 8 READMEs.
• Adds omlx: http://localhost:8000/v1/chat/completions to the endpoint preset code block in all 8 READMEs.
• Adds an omlx runtime-notes bullet (startup commands + default endpoint) only in README.md and README.zh-TW.md, matching the existing structure (the other 6 translations don't carry a Runtime notes section).

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
README.md	Adds omlx to features bullet, endpoint presets, and Runtime notes section.
README.zh-TW.md	Adds omlx to features bullet, endpoint presets, and Runtime notes section.
README.zh-CN.md	Adds omlx to features bullet and endpoint presets.
README.ja.md	Adds omlx to features bullet and endpoint presets.
README.ko.md	Adds omlx to features bullet and endpoint presets.
README.de.md	Adds omlx to features bullet and endpoint presets (ASCII-only spelling preserved).
README.es.md	Adds omlx to features bullet and endpoint presets (ASCII-only spelling preserved).
README.fr.md	Adds omlx to features bullet and endpoint presets (ASCII-only spelling preserved).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[linyiru]

English (en-US)

Thanks for the PR. This is a useful addition for users running local models on Apple Silicon.

I pushed a small follow-up commit to this PR branch to keep the documentation and Options UI aligned:

• Added omlx to the Local LLM endpoint preset dropdown.
• Added test coverage to confirm the omlx preset switches the provider to OpenAI Compatible and fills http://localhost:8000/v1/chat/completions.

Local validation passed:

• pnpm --filter @margin/extension exec vitest run src/options/providerSettings.test.ts src/options/optionsLayout.test.ts
• pnpm --filter @margin/extension check
• pnpm --filter @margin/extension check:extension

GitHub Actions currently show action_required because this is a fork PR and the workflows need maintainer approval before they run. That is not a test failure.

繁體中文

感謝提交這個 PR。這對使用 Apple Silicon 跑本機模型的使用者很有幫助。

我已經直接在這個 PR branch 補了一個小 commit，讓文件和 Options UI 保持一致：

• 在 Local LLM endpoint preset 下拉選單加入 omlx。
• 補上測試，確認選擇 omlx preset 會切換到 OpenAI Compatible，並填入 http://localhost:8000/v1/chat/completions。

本機驗證已通過：

• pnpm --filter @margin/extension exec vitest run src/options/providerSettings.test.ts src/options/optionsLayout.test.ts
• pnpm --filter @margin/extension check
• pnpm --filter @margin/extension check:extension

目前 GitHub Actions 顯示 action_required，是因為這是 fork PR，workflow 需要 maintainer 核准後才會執行；這不是測試失敗。

[linyiru]

@fredchu 感謝 PR，覺得溫馨

[fredchu]

@linyiru Lawrence 大大好久不見哈哈，發現優質專案來看看有沒有能貢獻的地方而已 🙈
本來想直接改 code，但想了想只是加一個 provider，好像加個範例就好了，不然以後可能會加不完 😂