DHT disagreement trips on a healthy grow: it compares tent, room and clone spans, which differ by design (lit tent +4 °C over a night room, domed clones +15 % R

Found on the dev bed (simulated rig) with ordinary values: 4×8 lit at 24.5 °C, room at night 20.3 °C, 2×4 21.4 °C / 73 % RH vs room 57 %. _dht_disagree_raw takes the span (max − min) across the tent, room and clone sensors and flags ≥ 4 °C or ≥ 15 % RH for 15 minutes as ‘DHT disagreement — check sensor placement’. Those three zones are separate environments by design: a lamp-lit tent routinely runs 3–6 °C over the room lung at night, and a clone tent runs well above room RH. The cue therefore fires nightly on a healthy rig, which trains the operator to ignore it (and, see the linked row, it is counted as CRITICAL). Fix idea: compare each sensor against the value the model expects for its zone (e.g. tent minus room given lamp state and exhaust), or only compare co-located pairs, or scale thresholds by lamp state; keep it a cue.