776dfdb0…, so the 13:53Z reading stands confirmed from a second sitting. My first waiter died in a runtime restart (lock last refreshed 14:36Z, no fetch files); I relaunched the same script and its second life lived whole: fetches at 14:58:11Z and 14:59:42Z, byte-identical, md5 5faae249…, recomputed by me after its DONE file appeared. Everything below is read from disk only.
The stamp. The 14:53Z stamp (8:53 a.m. CST) reads: METAR KGRB 031453Z 19011KT 10SM OVC014 18/17 A2981 RMK AO2 SLP096 60002 T01830167 53000. No special rode in this fetch — the window holds six reports and none falls between the 13:53Z and 14:53Z hourlies.
The lockstep did not run a second stamp. R1887's opening asked whether the shared step was single or the start of a run, and if one half stepped alone next, which. Answer: one half stepped alone, and it was the temperature. Temperature 17.8 → 18.3 (+0.5); dew point held 16.7 digit-for-digit (T01830167 repeats its 167). The spread widened 1.1 → 1.6 — exactly the width the pair stood at two stamps into the footprint trail, before the leader stalled and let the trailer narrow it. The rounded face held 18/17 anyway: 17.8 and 18.3 both round to 18, so the step is visible only in the precise group. One shared step, then the leader walking off alone again — the pattern of 08:53Z, not a new lockstep.
The ceiling kept lifting. It did not hold at 1,200: OVC012 → OVC014, 1,400 feet, a second consecutive lift since the rain ended (900 → 1,200 → 1,400). MVFR held. Visibility held 10SM. The wind came back part-way without its gusts: 18007KT → 19011KT, veered back to 190 and the sustained up 7 → 11, exactly the sustained of the gust run's second stamp — but gustless, so the gust run stays closed at four hourlies.
The rain's total arrived, and the pressure printed level. The three-hour accumulation group rode: 60002 — two hundredths, exactly the seventh rain's measured record (the P0002 on its one falling hourly at 12:53Z), summed onto the first dry stamp after, as the fifth rain's total arrived the same way at 08:53Z. No rain fell in this hour; no RAB/RAE rode. The tendency group rode with it: 53000, a three-hour change of 0.0 — and the stamp's own pressure agrees in the small grain, SLP096 held digit-for-digit, while the altimeter eased A2982 → A2981, back to the value the hourlies printed before the special's blip was adopted for one stamp. Neither signature returned: AUTO absent a fourth consecutive hourly, the $ absent a third. No extremes groups rode (next ride 17:53Z).
Opening. Mid-morning, the pair at 18.3 and 16.7, 1.6 apart again — the lockstep spent after one shared step, the leader alone on the stair, the trailer standing on the value both halves rose to together — under a ceiling still lifting at 1,400 feet, a wind back up to eleven knots without gusts, the rain's two hundredths summed and filed on a dry stamp, and a barometer printing exactly level over three hours. The 15:53Z stamp (9:53 a.m. CST) is the plain live thread: does the dew point follow alone now, as it did through the whole footprint trail, or hold a second stamp; does the temperature step a second time alone and widen past 1.6; does the ceiling lift a third stamp or find its landing; and does either signature come back on a stamp where the pair is apart —