autotest: Add catapult test

[Georacer]

An autotest for a catapult launch has been added for Plane.
An auxiliary set_servo method has been added to vehicle_test_suite.py.

[Georacer]

The way to call the test is

./Tools/autotest/autotest.py --no-clean --frame plane-catapult --speedup 10 build.Plane test.Plane.CatapultTakeoff

This PR is WIP because it's premature and failing.

There are two issues with it:

1. This test is specific to the plane-catapult frame. I'd like some advice on how to prevent it from running automatically for all frames.

2. For some reason the catapult refuses to launch the vehicle. I seem to successfully send the RC trigger to the catapult, but it does not produce forward acceleration.
   At some point the throttle picks up a bit (I don't know why) and the plane starts crawling on the ground).

Even more weirdly, autotest things the airspeed has risen, when the logs shows it hasn't.

AT-0004.1: airspeed=0.40 (want between 21.0 and 100)
AT-0004.1: AP: Triggered AUTO. GPS speed = 0.8
AT-0004.2: airspeed=13.59 (want between 21.0 and 100)
AT-0004.2: AP: SIM Hit ground at 3.284683 m/s
AT-0004.3: airspeed=27.58 (got between 21.0 and 100) (maintain=0.6/5.0)
AT-0004.3: AP: SIM Hit ground at 4.907945 m/s
AT-0004.3: AP: Holding course 35398 at 28.8m/s (-0.0)
AT-0004.4: airspeed=26.24 (got between 21.0 and 100) (maintain=1.7/5.0)
AT-0004.4: AP: SIM Hit ground at 4.911546 m/s
AT-0004.5: airspeed=25.58 (got between 21.0 and 100) (maintain=2.8/5.0)
AT-0004.6: AP: SIM Hit ground at 3.960280 m/s
AT-0004.6: airspeed=25.28 (got between 21.0 and 100) (maintain=3.8/5.0)
AT-0004.7: AP: SIM Hit ground at 3.963535 m/s
AT-0004.8: airspeed=25.92 (got between 21.0 and 100) (maintain=4.9/5.0)
AT-0004.8: AP: SIM Hit ground at 3.964641 m/s
AT-0004.8: Attained airspeed=25.871821010813992

There's a slight chance I'm looking at the wrong .bin. I inspected buildlogs/*2.bin and *3.bin.

[Georacer]

@peterbarker thanks for the suggestions. I've implemented it and it's running now.

Question:
I modified the basic.txt mission, that is used in FlyEachFrame. It contains a DO_LAND_START and more waypoints than I need for this test.
Should I use it regardless, or next time I should code a mission small enough for my needs?

Remark:
I've places this test in the known-failing list, as I believe this behaviour is now bugged.
The throttle never recedes during climb and the aircraft overspeeds:

If you agree, we can consider this PR done and I can open an Issue ticket.

[peterbarker]

Suggested change

	# Throw the catapult.
	self.progress("Throw the catapult")

Many times when I've put a comment like this in I've found it useful to put it in as a progress text instead. Just a suggestion.

[Georacer]

Thanks! For now it's pretty evident (perhaps even more useful) by plotting RCOU[7]. But I'll keep that in mind.

[peterbarker]

Question: I modified the basic.txt mission, that is used in FlyEachFrame. It contains a DO_LAND_START and more waypoints than I need for this test. Should I use it regardless, or next time I should code a mission small enough for my needs?

Whatever is most convenient, really. Crafting your own mission gives you exactly what you want with no complication, but takes longer. Note you can also load a "generic mission", a relatively new concept in the suite I'm still playing around with. e.g.

        self.load_generic_mission("CMAC-circuit.txt", strict=False)

There's something else you might be able to get away with, depending. Which is to just use "takeoff mode" to the appropriate altitude (it uses TECS). This may or may not reproduce, we do a few little interesting things in takeoff mode which might cause the problem to not reproduce.

You could possibly do something like this:

        self.takeoff(10)
        loc = self.mav.location()
        self.location_offset_ne(loc, 500, 500)
        loc.alt = 5000
        self.do_reposition(loc)

If you agree, we can consider this PR done and I can open an Issue ticket.

Yes - and you should use that ticket number in the disabled_tests method so people can look at the issue/analysis.

[Georacer]

I have updated the PR to switch from a check for receding throttle to a check for correct airspeed regulation.
This is because currently TECS assumes that the TAKEOFF happens at full throttle.

peter is a happy camper

[tridge]

closing in favor of the test code in #27174