Pot numeric values are re-recorded on every ~10 s poll although the probe only refreshes every 60 s – ~6 duplicate rows per reading per metric, while values tha

FIXED in PR #259: every seat’s numerics go through record_history_throttled with a 60 s heartbeat (RECORD_HEARTBEAT_S in esphome_client), so an unchanged probe value lands once a minute at most even though the ingest now snapshots every 2 s; Sonoff relay_on stays unconditional for the runtime ledger. Verify: SELECT COUNT(*) FROM fleet_history WHERE seat_id=’pot2′ AND metric=’moisture_pct’ AND ts > now-3600 is <= ~60 on a flat probe (was ~360). --- Original: the pot branch wrote every numeric unconditionally each poll; with the modbus probe on a 60 s interval the same reading landed 5-6 times.