ci: My Mesh card E2E fails on master since #456 (shard 3/3 timeout) #462

Problem

Since master a9562a6b (merge of #456), the Playwright shard 3/3 test "home My Mesh card: direct names, relay count, dialog lists both" has failed in both the original run and the rerun of master CI run 38713612837:

locator.waitFor: Timeout 15000ms exceeded.

Earlier master commits were green: 0f9a04a4, a768fd0a (#452) and 6f613a2b (#455). #456 changed how observer positions are resolved: GetNodePositionsByKeys, packet-path nodeByName/pubkey resolution, the position source, and rejection of 0,0 and invalid coordinates. The card comes from #451 (#76: Heard By split into direct and relay).

Task

1. Reproduce locally on a9562a6b and on 6f613a2b with the CI fixture and seeds. Run it several times to separate a regression from flakiness.
2. If it is a regression, find the root cause. Do not paper over it by raising the timeout. Fix it in the smallest correct place. Any change to Real observer positions before IATA fallback in geographic analytics #456 behaviour must keep its tests and guarantees: real position before IATA, and 0,0 and invalid coordinates rejected.
3. If it is flakiness, find the race and fix the test or the code. Do not add retries.

Acceptance criteria

The test passes reliably (5/5 locally) on the fixed head, and master CI is green again.