feat(yield): replace hardcoded multipliers with XGBoost regression model

[parinaB]

What does this PR do?

Replaces hardcoded yield multipliers in services/yield_service.py with a trained XGBoost regression model that learns yield relationships from growth stage, disease severity, weather conditions, and field size making predictions dynamic and realistic instead of static.

Related Issue

Closes #708

Changes made

• Added training/train_yield_model.py -->generates 3000 synthetic samples per crop using ICAR domain ranges, trains crop-specific XGBoost regressors with 5-fold cross-validation, saves models to ai_models/
• Added training/yield_model_config.json --> centralised domain config (base yields, disease severities, weather thresholds) sourced from ICAR-CICR data
• Updated services/yield_service.py --> loads trained model at inference time, builds feature vector with 3 interaction features (disease_impact, temp_humidity_stress, stage_health_interact), gracefully falls back to rule-based estimation if model file is missing
• Updated app.py --> passes crop_type from request into analyze_image() and estimate_yield() for all routes (/analyze, /api/analyze, /demo)
• Added test_yield.py --> validates XGBoost path, tomato/potato inference, and legacy fallback behaviour

Model Performance

Crop | CV MAE | CV R² | Samples -- | -- | -- | -- Cotton | 34.64 kg/acre | 0.89 | 3000 Tomato | 669.81 kg/acre | 0.78 | 3000 Potato | 462.21 kg/acre | 0.88 | 3000

SCREEENSHOT

Checklist

• I have tested my changes
• My code follows project guidelines
• No new errors introduced

[github-actions]

Thanks for creating a PR for your Issue! ☺️

We'll review it as soon as possible.
In the meantime, please double-check the file changes and ensure that all commits are accurate.

If there are any unresolved review comments, feel free to resolve them. 🙌🏼