MAESTRO: Merge PR #693 - prevent infinite restart loop that causes runaway API costs

Add restart limit (max 3 consecutive restarts) with exponential backoff
to prevent infinite generator restart loops. Also add defensive
memorySessionId checks in GeminiAgent and OpenRouterAgent before
expensive LLM calls to fail fast when session ID hasn't been captured.

Based on PR #693 by @ajbmachon (applied to current main).

Co-Authored-By: Andre Machon <ajbmachon2@gmail.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/services/worker/OpenRouterAgent.ts

@@ -150,6 +150,12 @@ export class OpenRouterAgent {
             session.lastPromptNumber = message.prompt_number;
           }
 
+          // CRITICAL: Check memorySessionId BEFORE making expensive LLM call
+          // This prevents wasting tokens when we won't be able to store the result anyway
+          if (!session.memorySessionId) {
+            throw new Error('Cannot process observations: memorySessionId not yet captured. This session may need to be reinitialized.');
+          }
+
           // Build observation prompt
           const obsPrompt = buildObservationPrompt({
             id: 0,
@@ -188,6 +194,11 @@ export class OpenRouterAgent {
           );
 
         } else if (message.type === 'summarize') {
+          // CRITICAL: Check memorySessionId BEFORE making expensive LLM call
+          if (!session.memorySessionId) {
+            throw new Error('Cannot process summary: memorySessionId not yet captured. This session may need to be reinitialized.');
+          }
+
           // Build summary prompt
           const summaryPrompt = buildSummaryPrompt({
             id: session.sessionDbId,