Fix AI assignment returning nothing: cap tokens, optimize prompt, show errors

- Cap maxTokens at 12000 (was unlimited dynamic calc that could exceed model limits)
- Replace massive EXISTING array with compact CURRENT_JUROR_LOAD counts and
  ALREADY_ASSIGNED per-project map (keeps prompt small across batches)
- Add coverage gap-filler: algorithmically fills projects below required reviews
- Show error state inline on page when AI fails (red banner with message)
- Add server-side logging for debugging assignment flow
- Reduce batch size to 10 projects

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

src/server/routers/roundAssignment.ts

@@ -152,6 +152,7 @@ export const roundAssignmentRouter = router({
       }
 
       // Call AI service
+      console.log(`[AI Assignment Router] Starting for ${projects.length} projects, ${jurors.length} jurors, ${input.requiredReviews} reviews/project, max ${maxPerJuror}/juror`)
       const result = await generateAIAssignments(
         jurors,
         projects,
@@ -159,6 +160,7 @@ export const roundAssignmentRouter = router({
         ctx.user.id,
         input.roundId,
       )
+      console.log(`[AI Assignment Router] Got ${result.suggestions.length} suggestions, success=${result.success}, fallback=${result.fallbackUsed}`)
 
       // Filter out COI pairs and already-assigned pairs
       const existingPairSet = new Set(existingAssignments.map((a) => `${a.userId}:${a.projectId}`))