re(flight): both axes at settled speeds -- shape confirmed, absolute scale is not

Burst design repeated with the HUD clock bracketed (01:01.86 -> 02:10.61, ~x1.26)
and roll measured in the same flight.

    pitch   87.0 @ 85    82.8 @ 308    47.0 @ 1077   min:max 1.85
    roll   125.0 @ 76   119.0 @ 291    82.6 @ 1023   min:max 1.51

SHAPE CONFIRMED on both axes: the rate interpolates between _Min (at minimum
speed) and _Max (at maximum speed), matching to 1.6% for pitch (vs PitchMinus
75/40 = 1.88) and 5.6% for roll (vs 200/125 = 1.60). Both ratios are
clock-independent so they stand regardless of the conversion, and pitch's
re-confirms ly+ = pitch-MINUS (reproducing 1.82 from the previous run).

ABSOLUTE SCALE DOES NOT MATCH, and not in the same direction:

    pitch  measured/predicted = 1.15, 1.21, 1.07   consistently OVER
    roll                      = 0.62, 0.64, 0.60   consistently UNDER, very flat

A clock error cannot explain this -- it would move both axes together and these
go opposite -- so the discrepancy is per-axis. Candidates, neither measured: a 1 s
burst not completing the angular-acceleration ramp (would under-read; fits roll's
flat 0.62, not pitch's excess), or an unidentified per-axis multiplier. Left as an
open question with the separating test named: 1 s versus 3 s bursts at the same
throttle, where the ramp hypothesis predicts the longer burst reads higher.

Also noted: the roll run correctly REFUSED on a WEAK row pin after the pitch
bursts left the craft tumbled -- and roll is immune to that labelling, which the
guard's own message says, so ALLOW_WEAK_PIN=1 is its documented exception. A guard
that states its own exception is worth having.

docs/re/captures/roll-burst-final.csv

@@ -0,0 +1,7 @@
+throttle,rep,settled_speed,burst_speed,rate_deg_per_wall_s
+min,0,103.5,99.3,165.34
+min,1,123.8,91.9,149.72
+cruise,0,434.7,352.0,149.9
+cruise,1,423.1,382.5,150.07
+max,0,1230.6,1258.2,102.76
+max,1,1624.8,1319.1,105.38