test(integration): wait for Tari to resync in the subnet phase (#201)

The moved-subnet phase does a full down->up, restarting Tari, then asserts
'tari synced (required)' via assert_running_state — but unlike the per-scenario
deploy path (run_scenario) it never waits for Tari to close its post-restart
offline gap. Tari re-peers over Tor slowly, so the assertion catches it at
'loading' and the phase flakes (seen on the v1.6.0 release e2e: 387 passed, the
sole failure 'tari synced expected [done] got [loading]'; monerod synced fine).

Mirror run_scenario's guard: wait_tari_synced 300 before the battery when
dashboard.tari_required is true. Product behaviour is unchanged — Tari always
syncs, the test just asserted too early.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: VijitSingh97

tests/integration/run.sh

@@ -1724,6 +1724,11 @@ run_subnet_scenario() {
     wait_monero_synced 120 || true
     wait_miner_running 180 || true
     wait_hashes_flowing 300 || true
+    # The down/up restarted Tari too, so — like the per-scenario deploy path — let it close its
+    # post-restart offline gap before assert_running_state's sync check, or we catch it mid-"loading".
+    if [ "$(jq_get "$config" '.dashboard.tari_required')" = "true" ]; then
+        wait_tari_synced 300 || true
+    fi
 
     # Subnet-specific live checks + the standard running-state battery on the moved subnet.
     assert_subnet_live "$config"