Epic: static fusion of re-exporter resource chains (ADR-0)

[avrabe]

Parent

`safety/adr/ADR-0.md` enumerates seven candidate design paths (D, E, F, G, H, I, J) for fixing the three failing wit-bindgen re-exporter fixtures (`resource_floats`, `resource_with_lists`, `resource-import-and-export`).

Pipeline

This epic drives the `scripts/explore/` pipeline (sibling to `scripts/mythos/`):

1. Rank the seven paths by viability × locus-of-fix
2. Prototype in parallel — one agent per viable path, each on a `feat/explore-` branch
3. Validate each prototype with a fresh agent
4. Emit `safety/adr/ADR-N.md` siblings under ADR-0 for each confirmed prototype
5. Synthesize — one agent reads all siblings, recommends one as `accepted`

Sub-issues (will be filed once rank pass completes)

Each path gets its own issue. The rank pass decides priority/parallelism.

Acceptance

• One ADR sibling promoted to `accepted` (humans, not the pipeline)
• The accepted path's prototype merged to main with the gating fixture flipped to a passing `runtime_test!`
• Other ADR siblings moved to `rejected` / `superseded` with reasons
• If the accepted path's locus is `wit-bindgen` or `spec`, an upstream PR / RFC tracking issue is also filed

Related

• Issue feat: per-component resource handle tables for 3-component chains #69 — feat: per-component resource handle tables for 3-component chains
• Issue 3-component re-exporter resource chains: implement shim module approach #92 — 3-component re-exporter resource chains: implement shim module approach
• PR docs: memo on wasmtime FACT reusability for Christof Petig #105 — docs/wasmtime-fact-reusability.md (related: shareable canonical-ABI emission)