fix: main's red Code map (97 stale pins from #1113) and Coverage (real-connection payment race)

[cryptskii]

main has two red checks since #1112 and #1113. This PR fixes both.

Code map: 97 stale pins (since #1113, 8017c77e3)

#1113 changed code under 97 pins whose evidence had been sealed before it. All 97 are code-class changes (production or evidence-test closures); none is semantic.

• Re-pinned with ci/intent_pins.py repin, one key at a time, at 8017c77e3.
• The map used is CI's code map for 2c8fe4b59, whose tree is main's exactly (3f1358ee23e8).
• The 71 evidence tests those pins name passed on that tree first: 6 cargo runs, 0 failed.
• make requirement-map-intent MAP=… INTENT_BUILT=android,node: 0 failing rows, 0 failing pins, 635 pinned.

Coverage: real_connection.rs:845 (since #1112, 7e084d713)

a_game_and_a_wallet_connect_over_the_real_relay asserted that the payment's outcome was CarriedOut and got Unspecified.

The cause is in the test, not the product:

• The game's account establishes the Paid fact from the transfer it accepted; connect.app.status takes in the inbox first.
• The wallet's answer reaches the game separately, over the relay.
• outcome is Unspecified exactly while no answer has arrived (answered: answer.is_some() in connect_routes).

The test waited for the fact only. It now waits for the answer as well before reading the outcome. Only the Coverage job, an instrumented debug build that runs on pushes to main and not on PRs, was slow enough to hit it.

Verified locally: the test passes (117 s), the real-code guard is clean, and rustfmt is clean. This file is in no pin's evidence or closure, so the re-pin above still holds.