Added CICIDS dataset and updated training pipeline for anomaly detection

[himelds]

Summary

• Added a cleaned subset of the CICIDS2017 dataset in datasets
• Selected relevant numeric features for anomaly detection
• Removed missing and infinite values

Model Training

• Updated training script to load dataset instead of hardcoded data
• Trained Isolation Forest model
• Saved trained model in models/ directory

Result

The training pipeline now runs successfully with real data and produces a usable anomaly detection model for security experimentation.

Note: Training script saves the trained model to the models/ directory (excluded via .gitignore)