[C2] Channel-contention test

[radical]

Track C validation. Depends on B3 (channel scope fix).

Test the channel cross-contamination fix:

1. Simulate route A writing channel=daily, route B reading. Assert correct semantics per Q3 decision.
2. Concurrent aspire update --channel X from two routes — last writer wins, no corruption, file always parseable.
3. PR install sets channel=pr-N, gets uninstalled, winget binary creates project — must NOT see pr-N channel (W34, W35 regression).

Scenarios covered

S4.1–S4.4.

---

Part of: #29