feat: add llmman as an LLM provider

[ericcurtin]

Adds llmman, a local model runner that serves the Ollama API (alongside OpenAI- and Anthropic-compatible ones) on port 17434.

Because the wire protocol is the same, the provider subclasses the existing Ollama implementation and overrides only the provider name and default apiBase — 18 lines instead of duplicating 833. Model listing, FIM, embeddings and streaming all carry over.

Two other touchpoints:

• toolSupport delegates to the same name-based heuristic, since llmman serves the same local models.
• fetchModels needs no case of its own — the default branch goes through listModels, which the inherited implementation answers from the running server rather than scraping a remote catalog. That's the correct behaviour here, not an omission.

Testing: tsc --noEmit -p core/tsconfig.json reports 157 errors both before and after this change (all pre-existing "cannot find module" for unbuilt workspace packages like @continuedev/config-yaml), so nothing new is introduced. I couldn't run the extension end-to-end against a live server, so that part is unverified.

AI-assisted, reviewed before submitting.

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}