feat(ui): add prediction state management and PredictionPanel component

- Create prediction type definitions in src/types/predictions.ts
- Add prediction state management to page.tsx with caching
- Implement PredictionPanel component with:
  - Master visibility toggle
  - Model info display (name, version, type, metrics)
  - Action buttons (Run on Visible, Predict All)
  - Confidence threshold slider
  - Label filter checkboxes with per-class metrics
  - Disagreement filter toggle
  - Prediction summary display
  - Model server offline banner
- Add on-demand and batch prediction fetching
- Implement prediction caching by chart and model version
- Add health polling for inference API (30s interval when offline)
- Ensure annotation tools work independently of prediction API

Tasks completed: 9.1-9.5, 12.1-12.3 (59/78 total)

src/types/predictions.ts

@@ -0,0 +1,118 @@
+/**
+ * Prediction types for ML model inference
+ */
+
+export interface PredictionSpan {
+  label: string;
+  start_time: number;
+  end_time: number;
+  avg_confidence: number;
+  candle_count: number;
+}
+
+export interface PerCandlePrediction {
+  time: number;
+  label: string;
+  confidence: number;
+}
+
+export interface ModelInfo {
+  model_name: string;
+  model_version: string;
+  model_type: string;
+  experiment_name: string;
+  run_id: string;
+  trained_at: string;
+  feature_count: number;
+  label_names: string[];
+}
+
+export interface PerClassMetrics {
+  [label: string]: {
+    precision: number;
+    recall: number;
+    f1_score: number;
+    support: number;
+  };
+}
+
+export interface ModelMetrics {
+  accuracy: number;
+  f1_macro: number;
+  f1_weighted: number;
+  per_class: PerClassMetrics;
+}
+
+export interface ModelInfoResponse {
+  model_info: ModelInfo;
+  metrics: ModelMetrics;
+  label_config: {
+    name: string;
+    color: string;
+  }[];
+}
+
+export interface PredictRequest {
+  candles: {
+    time: number;
+    open: number;
+    high: number;
+    low: number;
+    close: number;
+    volume?: number;
+  }[];
+}
+
+export interface PredictResponse {
+  predictions: PerCandlePrediction[];
+  spans: PredictionSpan[];
+  model_info: {
+    model_name: string;
+    model_version: string;
+  };
+}
+
+export interface BatchPredictRequest {
+  pair: string;
+  timeframe: string;
+  start_time: number;
+  end_time: number;
+  batch_size?: number;
+}
+
+export interface PredictionState {
+  spans: PredictionSpan[];
+  perCandlePredictions: PerCandlePrediction[];
+  isLoading: boolean;
+  error: string | null;
+  modelInfo: ModelInfoResponse | null;
+  visible: boolean;
+  confidenceThreshold: number;
+  selectedLabels: Set<string>;
+  autoPredict: boolean;
+  cacheKey: string | null;
+}
+
+export type DisagreementType = 
+  | 'missed_by_model'    // Human annotation but no prediction
+  | 'missed_by_human'    // Prediction but no human annotation
+  | 'label_mismatch';    // Both present but different labels
+
+export interface Disagreement {
+  type: DisagreementType;
+  humanSpan?: {
+    id: number;
+    label: string;
+    start_time: number;
+    end_time: number;
+  };
+  predictionSpan?: PredictionSpan;
+  overlap_ratio?: number;
+}
+
+export interface PredictionSummary {
+  total_predictions: number;
+  total_human_annotations: number;
+  agreements: number;
+  disagreements: Disagreement[];
+}