#175: Add type-checked Pydantic training config

[fadee26]

Summary of Changes

This PR addresses three issues: #175, #182, #190

🔧 #175: Add type-checked API for training config

• New file astroml/training/config.py with Pydantic models
• Models: EarlyStoppingConfig, TemporalSplitConfig, OptimizerConfig, TrainingConfig
• TrainingConfig includes from_yaml() and to_yaml() methods

📊 #182: Ingestion backfill idempotency

• Added ProcessedLedger model in astroml/db/schema.py to track processed ledgers
• Updated preprocess_to_parquet with skip_processed=True parameter
• Skips ledgers marked as completed, and uses processing status for concurrency safety

⚡ #190: Feature extractor performance bottleneck

• Replaced offset/limit with keyset pagination in pipeline_structural_importance.py
• Fix missing Transaction import
• Keyset pagination avoids full table scans for large datasets

closes #190
closes #182
closes #175

[drips-wave]

@fadee26 Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits