PX4Accelerometer apply sensor rotation before scaling

[dagar]

• prevents loss of numerical precision
• fixes Accelerometer calibration problem on FMUV5 #11695

[dagar]

This is also worth noting for downstream of the sensor drivers where board rotations are applied on the scaled floating point data (SENS_BOARD_ROT, SENS_BOARD_{X, Y, Z}_OFF.

• sensors module
• calibration
• mc_att_control

https://github.com/PX4/Firmware/blob/f402f68fb19b96152486ef95e1fa2b5cbc91b39d/src/modules/sensors/voted_sensors_update.cpp#L135-L138

Switching to double precision floating point on newer platforms like px4_fmu-v5 would be one relatively easy solution.

[bresch]

Nice!

INFO  [ecl/validation] validator: best: 0, prev best: 0, failsafe: NO (0 events)
INFO  [ecl/validation] sensor #0, prio: 100, state: OK
INFO  [ecl/validation] 	val:  -0.0215, lp:  -0.0282 mean dev:  -0.0034 RMS:   0.0589 conf:   1.0000
INFO  [ecl/validation] 	val:   0.0272, lp:   0.0396 mean dev:  -0.0001 RMS:   0.0180 conf:   1.0000
INFO  [ecl/validation] 	val:  -9.8133, lp:  -9.8210 mean dev:   0.0008 RMS:   0.0385 conf:   1.0000
INFO  [ecl/validation] sensor #1, prio: 99, state: OK
INFO  [ecl/validation] 	val:  -0.1556, lp:  -0.1544 mean dev:  -0.0014 RMS:   0.0308 conf:   1.0000
INFO  [ecl/validation] 	val:   0.0821, lp:   0.0560 mean dev:  -0.0003 RMS:   0.0237 conf:   1.0000
INFO  [ecl/validation] 	val:  -9.7675, lp:  -9.7904 mean dev:   0.0002 RMS:   0.0456 conf:   1.0000

[LorenzMeier]

@bresch Could you please work with @dagar on looking at the other options to see if we miss out on the other locations?