Fix TestLightRestartTestSuite/TestLightNodeResumesSamplingAfterRestart in tastora

A lot of `ERROR module/p2p ... Host is not reachable from the public network`

Possible solutions:
Make the test independent of public-network reachability and flaky peer discovery by forcing local-only connectivity and seeding peers explicitly after restart.

1) In the failing restart test setup, disable strict reachability expectations

If the test currently waits for “publicly reachable” state (directly or indirectly), remove that as a success condition. In CI Docker networks, public reachability is commonly false and should not block DAS-resume assertions.

2) Seed at least one stable peer for the light node after restart

On restart, explicitly reconnect the light node to a known full/bridge peer address used in the test harness (instead of relying on DHT advertising/discovery timing).

Locate TestLightRestartTestSuite/TestLightNodeResumesSamplingAfterRestart and run the failing test to observe the reachability error. Inspect its restart setup and test harness peer configuration; done means the test does not require public-network reachability, reconnects to a known peer after restart, and passes its DAS-resume assertions.

Written by the indexing model from the issue text.