Closes#2407. FSK PCM, ~26 us bit width, inverted levels. Reverse
engineered from 3 years of prior investigation in the issue (confirmed
preamble, message framing, and CRC-16, though the actual sensor payload
was never solved) plus new work correlating the reporter's own log
against the one piece of ground truth available: display readings
drifting from 7.1 C/59% humidity to 9.9 C/49% humidity over a real
session.
Humidity is confirmed exactly against both ends of that session (not
approximately -- exact integer matches). Temperature is fit from the
same 2 reference points, which is the bare minimum for a 2 parameter
linear model with no slack to validate against a third point, so it's
documented as a rough estimate pending wider-range calibration data.
Only the "long" message kind (of two seen on air, sharing a 14 byte
header) is decoded; the "short" kind's extra payload (suspected wind
data) is not decoded.
4d37e6f9da