[WIP] Timestamps - Add timestamp chaining through timestamp_sample

[bresch]

The timestamp_sample field has been used to propagate the timestamp of a measurement through the actuator_controls topic. I completed the chain to have this working though the VTOL code in order to estimate the dead-time this module adds and being able to get the complete computation time at each loop.
Here is the result of a run on a PixHawk Pro autopilot from Drotek:

sensor_combined -> 0.75ms -> vehicle_attitude -> 0.25ms -> actuator_controls_virtual -> 3.25ms -> actuator_controls -> 0.5ms -> actuator_outputs

The total delay - from sensors_combined to actuator_outputs is approximately 4.75ms. Which makes a maximum loop frequency of ~211Hz... not 250.

@dagar Not sure if you would like to merge this, but at least you have the results here.

[dagar]

Thanks! At a minimum I'll do some testing across a range of different configurations and then we'll see if it's worth having this in all the time. My only hesitation is increasing the message size for something that we aren't regularly going to be looking at.

[dagar]

Added [WIP] to block merging.

[dagar]

I'm not sure about this. There are other actuator_controls publications in the system that might be throwing this off. https://github.com/PX4/Firmware/blob/master/src/modules/fw_att_control/fw_att_control_main.cpp#L1233

For now only using actuator_controls_0 should be sufficient if in the VTOL case you always put the timestamp_sample coming from the active controller.

[dagar]

@bresch out of curiosity did you try in both MC and FW VTOL modes?

[bresch]

@dagar Only MC and it was disarmed.

[dagar]

Continued in #9388 (partially).