AI-powered assignment generation with enriched data and streaming UI

- Add aiPreview mutation with full project/juror data (bios, descriptions,
  documents, categories, ocean issues, countries, team sizes)
- Increase AI description limit from 300 to 2000 chars for richer context
- Update GPT system prompt to use all available data fields
- Add mode toggle (AI default / Algorithm fallback) in assignment preview
- Lift AI mutation to parent page for background generation persistence
- Show visual indicator on page while AI generates (spinner + progress card)
- Toast notification with "Review" action when AI completes
- Staggered reveal animation for assignment results (streaming feel)
- Fix assignment balance with dynamic penalty (25pts per existing assignment)

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

src/server/services/ai-assignment.ts

@@ -35,11 +35,15 @@ const ASSIGNMENT_BATCH_SIZE = 15
 const ASSIGNMENT_SYSTEM_PROMPT = `You are an expert jury assignment optimizer for an ocean conservation competition.
 
 ## Your Role
-Match jurors to projects based on expertise alignment, workload balance, and coverage requirements.
+Match jurors to projects based on expertise alignment, workload balance, geographic diversity, and coverage requirements. You have access to rich data about both jurors and projects — use ALL available information to make optimal assignments.
+
+## Available Data
+- **Jurors**: expertiseTags (areas of expertise), bio (background description with deeper domain knowledge), country, currentAssignmentCount, maxAssignments
+- **Projects**: title, description (detailed project overview), tags (with confidence 0-1), category (e.g. STARTUP, BUSINESS_CONCEPT), oceanIssue (focus area like CORAL_REEFS, POLLUTION), country, institution, teamSize, fileTypes (submitted document types)
 
 ## Matching Criteria (Weighted)
-- Expertise Match (50%): How well juror tags/expertise align with project topics. Project tags include a confidence score (0-1) — weight higher-confidence tags more heavily as they are more reliably assigned. A tag with confidence 0.9 is a strong signal; one with 0.5 is uncertain.
-- Workload Balance (30%): Distribute assignments evenly; prefer jurors below capacity
+- Expertise & Domain Match (50%): How well juror tags, bio, and background align with project topics, category, ocean issue, and description. Use bio text to identify deeper domain expertise beyond explicit tags — e.g., a bio mentioning "20 years of coral research" matches coral-related projects even without explicit tags. Weight higher-confidence tags more heavily.
+- Workload Balance (30%): Distribute assignments as evenly as possible; strongly prefer jurors below capacity. Never let one juror get significantly more assignments than another.
 - Minimum Target (20%): Prioritize jurors who haven't reached their minimum assignment count
 
 ## Output Format
@@ -51,18 +55,20 @@ Return a JSON object:
       "project_id": "PROJECT_001",
       "confidence_score": 0.0-1.0,
       "expertise_match_score": 0.0-1.0,
-      "reasoning": "1-2 sentence justification"
+      "reasoning": "1-2 sentence justification referencing specific expertise matches"
     }
   ]
 }
 
 ## Guidelines
-- Each project should receive the required number of reviews
+- Each project MUST receive the required number of reviews — ensure full coverage
+- Distribute assignments as evenly as possible across all jurors
 - Do not assign jurors who are at or above their capacity
-- Favor geographic and disciplinary diversity in assignments
-- confidence_score reflects overall assignment quality; expertise_match_score reflects tag overlap only
-- A strong match: shared expertise tags + available capacity + under minimum target
-- An acceptable match: related domain + available capacity
+- Favor geographic diversity: avoid assigning jurors from the same country as the project when possible
+- Consider disciplinary diversity: mix different expertise backgrounds per project
+- confidence_score reflects overall assignment quality; expertise_match_score reflects tag/expertise overlap
+- A strong match: shared expertise tags + relevant bio background + available capacity
+- An acceptable match: related domain/ocean issue + available capacity
 - A poor match: no expertise overlap, only assigned for coverage`
 
 // ─── Types ───────────────────────────────────────────────────────────────────
@@ -88,6 +94,8 @@ interface JurorForAssignment {
   name?: string | null
   email: string
   expertiseTags: string[]
+  bio?: string | null
+  country?: string | null
   maxAssignments?: number | null
   _count?: {
     assignments: number
@@ -101,6 +109,12 @@ interface ProjectForAssignment {
   tags: string[]
   tagConfidences?: Array<{ name: string; confidence: number }>
   teamName?: string | null
+  competitionCategory?: string | null
+  oceanIssue?: string | null
+  country?: string | null
+  institution?: string | null
+  teamSize?: number
+  fileTypes?: string[]
   _count?: {
     assignments: number
   }