Add the Unsloth warm-fork demo (one golden, three concurrent fine-tunes)

[BinSquare]

Runnable demo + captured output: load Qwen2.5-1.5B once in a frozen golden, fork three clones that each fine-tune a different skill concurrently on one 8 GB GPU.

---

[stage-review]

Ready to review this PR? Stage has broken it down into 3 individual chapters for you:

	Title
1	Document the Unsloth warm-fork architecture
2	Implement the multi-task training workload script
3	Add the demo orchestration script

_{Chapters generated by Stage for commit 6893a09 on Jul 15, 2026 8:37pm UTC.}