feat: add LM Studio headless server support

[vlordier]

Summary

• Add LM Studio runtime configuration to _models/config.yaml
• Add example model config for LM Studio (default port 1234)
• Add unit tests for LM Studio model and runtime config

Changes

• examples/localcowork/_models/config.yaml: Added lmstudio runtime and model
• examples/localcowork/src-tauri/src/inference/config.rs: Added 2 tests
• examples/localcowork/src-tauri/src/inference/client.rs: Added 1 test + fixture

Testing

• All 363 Rust unit tests pass
• Clippy clean

Note

LM Studio must be started manually from the app GUI (no CLI for headless mode).

[Copilot]

Pull request overview

Adds configuration and test scaffolding to support using an LM Studio OpenAI-compatible local server (default http://localhost:1234/v1) within the LocalCowork example app.

Changes:

• Added LM Studio model + runtime entries to the LocalCowork YAML configs.
• Added unit tests to ensure LM Studio model configs deserialize and can be selected by the inference client.
• Added a new top-level examples/localcowork/config.yaml config file (in addition to _models/config.yaml).

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
examples/localcowork/src-tauri/src/inference/config.rs	Adds LM Studio-focused deserialization tests (and minor formatting changes).
examples/localcowork/src-tauri/src/inference/client.rs	Extends test fixture with an LM Studio model and adds a targeted constructor test.
examples/localcowork/config.yaml	Introduces a top-level YAML config containing models + runtimes + fallback chain.
examples/localcowork/_models/config.yaml	Adds LM Studio model and runtimes.lmstudio configuration; also includes formatting/tidy updates.

---

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

[Copilot]

The new runtimes.lmstudio section is currently ignored by the Rust loader because ModelsConfig has no runtimes field and Serde will drop unknown keys. If this is meant to drive runtime startup/health-check behavior, it should be added to ModelsConfig (and used) or the YAML should explicitly document that the runtimes map is informational-only so readers don’t assume it affects behavior.

[vlordier]

@copilot open a new pull request to apply changes based on this feedback

[Paulescu]

Thanks for the LM Studio addition. The core idea is useful, but the PR is too broad and stale to merge as-is. Could you rebase on current main and narrow the scope before we take another look?

What to keep:

• The new lmstudio-default model entry in _models/config.yaml.
• The new lmstudio block under the informational runtimes: section.
• The 3 unit tests in inference/client.rs and inference/config.rs.

What to drop:

• The whitespace edits that strip the double-space before YAML comments (: foo # bar → : foo # bar) across the rest of config.yaml. Unrelated to LM Studio.
• The LFM2-24B-A2B-Preview → LFM2-24B-A2B rename of display_name, model_path, and the Hugging Face link. Main has since moved past Preview naming entirely (fix(localcowork): update model from gated Preview to public GGUF release #93, currently LFM2-24B-A2B-GGUF), so this hunk is now obsolete and will conflict on rebase.
• The cosmetic reformatting in inference/config.rs (multi-line assert!, wrapping the resolve_active_model signature).

One inconsistency to fix while you're in there: the PR description says "LM Studio must be started manually from the app GUI (no CLI for headless mode)", but the runtimes.lmstudio block specifies command: "lms", args: ["server", "start"]. The lms CLI does work headlessly, so the description is wrong; remove that line or update it.

Once the diff is rebased and scoped just to the LM Studio additions (model entry, runtime entry, tests), this should be a quick merge.