fix(tests): prevent OOM and infinite hangs in incident search loop

Three layered defenses against Chrome/Node.js OOM (exit 137) and
infinite hangs in findIncidentWithAlert's retry loop:

1. cy.reload() at start of each search iteration — releases browser
   DOM from previous iteration, preventing browser-side accumulation.

2. _quietSearch/_qLog() pattern — suppresses Cypress command logging
   (and DOM snapshot serialization, ~1-5 MB each) during search.
   Without this, ~40 commands * 15+ iterations = 600+ snapshots OOM.

3. Hard timeout safety net (35 min) — Date.now()-based kill switch
   that fires if cy.waitUntil's timeout breaks due to cy.reload()
   interfering with the Cypress command queue.

Also adds warmUpForPlugin() page object method for plugin loading
race condition (used by subsequent commit).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: Claude Agent

web/cypress/e2e/incidents/regression/03-04.reg_e2e_firing_alerts.cy.ts

@@ -42,6 +42,9 @@ describe(
     before(() => {
       cy.beforeBlockCOO(MCP, MP, { dashboards: false, troubleshootingPanel: false });
 
+      // Reset the search timeout so this spec gets a fresh 35-minute window
+      incidentsPage.resetSearchTimeout();
+
       cy.log('Create firing alert for testing');
       cy.cleanupIncidentPrometheusRules();
       cy.createKubePodCrashLoopingAlert().then((alertName) => {