AI Grouping of Similar Activity Types in Volume Distribution

[hdkiller]

💡 Summary

Implement AI-driven automatic grouping of similar activity types in the 'Volume Distribution' graph. Currently, activities from different sources (Garmin, Strava, Intervals.icu) can create 'duplicate' activity types with slightly different labels (e.g., 'E_bike' and 'ebikeride' for electric bike rides).

🤷 Motivation / Use Case

This makes the volume distribution chart less clear and accurate. Consolidating these labels would improve data clarity and user experience, especially until manual activity type editing is fully rolled out.

📋 Detailed Requirements

• AI-driven mapping of source-specific activity strings to canonical activity types.
• Aggregation of volume data based on these canonical groups in the Volume Distribution graph.
• Handling of common duplicates (e.g., E-Bike variations, specialized running/cycling types).

✅ Acceptance Criteria

• The Volume Distribution graph displays consolidated activity types.
• User does not see multiple entries for the same functional activity type (e.g., just one "E-Bike" instead of "E-Bike" and "ebikeride").
• System correctly handles edge cases from different integration sources.