fix(build): cap parallel compilation threads to prevent CPU lockup

[Katsoragi]

Summary

Caps compiler thread usage to (total_cores - 2) when compiling llama.cpp natively to avoid host system lockups. Previously, it used nproc (all cores), which froze user systems. It also handles local endpoint context window queries properly.

Target branch

• This PR targets dev, not main. All PRs land in dev; main is curated by the maintainer at each release. If your PR is on main by accident, click "Edit" on this PR and change the base.

Linked Issue

Fixes #5161

Type of Change

• Bug fix (non-breaking — fixes a confirmed issue)
• New feature (non-breaking — adds new behaviour)

Checklist

• I searched open issues and open PRs — this is not a duplicate.
• This PR targets dev
• My changes are limited to the scope described above — no unrelated refactors or whitespace changes mixed in.
• I actually ran the app and verified the change works end-to-end.

How to Test

Verify that llama.cpp native builds use the capped number of threads and do not freeze the host.

Checks Run

• Run pytest suite: pytest tests/test_model_defaults.py (All Passed).