Add the training zones tables from SYNTHESIS §7. Split from #20 (HR Zone Configuration) — the schema is significantly richer than that issue describes.

Tables

training_zones

• Zone definitions for HR, power, and pace (not just HR)
• Versioned with effective_from dates (GoldenCheetah pattern) — insert new rows, don't overwrite old ones. Historical activities resolve to zones active at the time.
• Both absolute and percentage-based: absolute values for coach-set or provider-synced zones, percentages for standard systems (Coggan, Karvonen, Friel)
• Per-sport overrides: sport_type = NULL = default, sport_type = 'cycling' = cycling-specific override
• TRIMP weighting on HR zones for HR-based training load calculation
• UNIQUE(athlete_id, zone_type, sport_type, zone_number, effective_from)

sport_zone_priorities

• Per-sport configuration: which metric drives time-in-zone and training load
• load_order and tiz_order (e.g., 'POWER_HR_PACE' for cycling)
• default_equipment_id for auto-assigning gear per sport

Zone resolution at query time

1. Look up (athlete_id, zone_type, sport_type, effective_from <= activity.start_date) ORDER BY effective_from DESC LIMIT 1
2. If percentage-based: resolve using nearest athlete_metrics anchor value (FTP, max HR, etc.)
3. If absolute: use directly
4. If no sport-specific zones: fall back to sport_type = NULL (default)

Depends on

Schema Migration

Blocks

#20, #21, #24, #25

Reference

See docs/api-research/SYNTHESIS.md §7 (training_zones table, sport_zone_priorities table).