fix: prevent phase stuck in_progress when merge fails

In YOLO mode, if mergePhaseIntoInitiative() threw (e.g., branch
doesn't exist, merge conflict), the error propagated and completePhase()
was never called — leaving the phase permanently stuck at in_progress.

Also wrap per-phase recovery in try-catch so one failing phase doesn't
abort the entire recoverDispatchQueues() loop.

apps/server/execution/orchestrator.ts

@@ -353,7 +353,11 @@ export class ExecutionOrchestrator {
     if (initiative.executionMode === 'yolo') {
       // Merge phase branch into initiative branch (only when branches exist)
       if (initiative.branch) {
-        await this.mergePhaseIntoInitiative(phaseId);
+        try {
+          await this.mergePhaseIntoInitiative(phaseId);
+        } catch (err) {
+          log.error({ phaseId, err: err instanceof Error ? err.message : String(err) }, 'phase merge failed, completing phase anyway');
+        }
       }
       await this.phaseDispatchManager.completePhase(phaseId);
 
@@ -663,6 +667,7 @@ export class ExecutionOrchestrator {
       const phases = await this.phaseRepository.findByInitiativeId(initiative.id);
 
       for (const phase of phases) {
+        try {
         // Re-queue approved phases into the phase dispatch queue
         if (phase.status === 'approved') {
           try {
@@ -734,6 +739,9 @@ export class ExecutionOrchestrator {
             phasesRecovered++;
           }
         }
+        } catch (err) {
+          log.error({ phaseId: phase.id, err: err instanceof Error ? err.message : String(err) }, 'phase recovery failed, continuing with remaining phases');
+        }
       }
     }