Fix multiple critical bugs in onboarding training step

This commit addresses all identified bugs and issues in the training code path:

## Critical Fixes:
- Add get_start_time() method to TrainingLogRepository and fix non-existent method call
- Remove duplicate training.started event from API endpoint (trainer publishes the accurate one)
- Add missing progress events for 80-100% range (85%, 92%, 94%) to eliminate progress "dead zone"

## High Priority Fixes:
- Fix division by zero risk in time estimation with double-check and max() safety
- Remove unreachable exception handler in training_operations.py
- Simplify WebSocket token refresh logic to only reconnect on actual user session changes

## Medium Priority Fixes:
- Fix auto-start training effect with useRef to prevent duplicate starts
- Add HTTP polling debounce delay (5s) to prevent race conditions with WebSocket
- Extract all magic numbers to centralized constants files:
  - Backend: services/training/app/core/training_constants.py
  - Frontend: frontend/src/constants/training.ts
- Standardize error logging with exc_info=True on critical errors

## Code Quality Improvements:
- All progress percentages now use named constants
- All timeouts and intervals now use named constants
- Improved code maintainability and readability
- Better separation of concerns

## Files Changed:
- Backend: training_service.py, trainer.py, training_events.py, progress_tracker.py
- Backend: training_operations.py, training_log_repository.py, training_constants.py (new)
- Frontend: training.ts (hooks), MLTrainingStep.tsx, training.ts (constants, new)

All training progress events now properly flow from 0% to 100% with no gaps.

services/training/app/ml/trainer.py

@@ -6,7 +6,7 @@ Main ML pipeline coordinator using repository pattern for data access and depend
 from typing import Dict, List, Any, Optional
 import pandas as pd
 import numpy as np
-from datetime import datetime
+from datetime import datetime, timezone
 import structlog
 import uuid
 import time
@@ -187,7 +187,10 @@ class EnhancedBakeryMLTrainer:
 
                 # Event 2: Data Analysis (20%)
                 # Recalculate time remaining based on elapsed time
-                elapsed_seconds = (datetime.now(timezone.utc) - repos['training_log']._get_start_time(job_id) if hasattr(repos['training_log'], '_get_start_time') else 0) or 0
+                start_time = await repos['training_log'].get_start_time(job_id)
+                elapsed_seconds = 0
+                if start_time:
+                    elapsed_seconds = int((datetime.now(timezone.utc) - start_time).total_seconds())
 
                 # Estimate remaining time: we've done ~20% of work (data analysis)
                 # Remaining 80% includes training all products
@@ -285,7 +288,8 @@ class EnhancedBakeryMLTrainer:
         except Exception as e:
             logger.error("Enhanced ML training pipeline failed",
                         job_id=job_id,
-                        error=str(e))
+                        error=str(e),
+                        exc_info=True)
 
             # Publish training failed event
             await publish_training_failed(job_id, tenant_id, str(e))
@@ -397,7 +401,8 @@ class EnhancedBakeryMLTrainer:
             logger.error("Single product model training failed",
                         job_id=job_id,
                         inventory_product_id=inventory_product_id,
-                        error=str(e))
+                        error=str(e),
+                        exc_info=True)
             raise
     
     def _serialize_scalers(self, scalers: Dict[str, Any]) -> Dict[str, Any]: