Generate Boatstack from the product engineering loop

[bigboateng]

Outcome

Makes Intelligence Flow the source of truth for Boatstack and adds a deterministic public-repository projection.

What changes

• generates Boatstack's evidence-led README, loop-engineering docs, worked example, tests, and skill from examples/12-product-engineering-loop
• brands thin Cursor, Codex, Claude, and GitHub adapters without changing the model-neutral workflow contract
• records every downstream file hash and source commit in UPSTREAM.json
• adds a Boatstack-owned polling workflow that proposes upstream changes by PR and never auto-merges
• refuses to overwrite manually modified generated files

Evidence

• python3 -m unittest discover -s examples/12-product-engineering-loop/tests -v — 7 passed
• generated Boatstack distribution tests — 3 passed
• canonical and generated skills pass quick_validate.py
• git diff --check passes

Boundary

Boatstack owns its generated presentation surface. Workflow semantics, gates, templates, and failure moves remain upstream in Intelligence Flow.