v0.1.2

Highlights

• Default ./run.sh build targets TheTom/llama-cpp-turboquant@feature/turboquant-kv-cache.
• ./run.sh warmup [model...] can load models early through llama-swap's upstream health route.
• Hugging Face -hf assets are documented as lazy downloads, with warmup and preload guidance.
• Rerank support is documented as a first-class route with qmd-rerank and /v1/rerank examples.
• config.yml.example is synced with the active QMD aliases: qmd-generate, qmd-embed, and qmd-rerank.
• README wording was cleaned up to keep the docs concise and approachable.

Included Commits

• 5a6368e Simplify README wording
• fc3f841 Document QMD rerank config and API
• 5ecc23e Add warmup flow and turboquant defaults