Tracker: wait for first command in guided mode before doing anything

[peterbarker]

without this it's hard to go into guided mode without the vehicle moving to extremes. We reject various commands in the mavlink layer if you're not in guided...

[peterbarker]

.... need to stop continuous yaw servos turning...

[github-actions]

Global Size Summary (Total Flash delta in bytes)

Built from bb63c69 rebased onto dc78bc5

Board	Periph	tracker	blimp	bootloader	copter	heli	iofw	plane	rover	sub
DrotekP3Pro	_	_	_	*	_	_	_	_	_	_
Durandal	_	+84	*	_	*	*	_	*	*	*
Hitec-Airspeed	*	_	_	_	_	_	_	_	_	_
KakuteF7	_	+80	*	_	*	*	_	*	*	*
MambaH743v4	_	+84	*	_	*	*	_	*	*	*
MatekF405-CAN	_	_	_	*	_	_	_	_	_	_
MatekF405	_	+80	*	_	*	*	_	*	*	*
MatekH743-bdshot	_	+84	*	_	*	*	_	*	*	*
Pixhawk1-1M	_	+84	*	_	*	*	_	*	*	*
disco	_	-20	*	_	*	*	_	*	*	*
f103-GPS	*	_	_	_	_	_	_	_	_	_

[AP-Review]

Automated review note — AI-generated (Claude+Codex), validated against the live diff. Please sanity-check before acting.
Verdict: REQUEST CHANGES

Reviewed at head bb63c6984c.
Full report: https://firmware.ardupilot.org/Tools/APReview/DevCallReviews/PRReviews/ardupilot/ardupilot/23525/1.html#pr23525

Thanks for picking this up again. Holding the current attitude on entry is a good fix for the jump to north and level.

There's one problem: ModeGuided::enter() doesn't reset the yaw and pitch PIDs (mode_guided.cpp:12). After SCAN, or AUTO with no vehicle, the controllers still hold the old error. On the first GUIDED loop the error drops to about zero, and the D term kicks the tracker about 5 deg before any command arrives. On blu6 your new GUIDEDHoldsAttitudeOnEntry test fails most runs at this head: 0/3 standalone, 2/7 in the full suite. With tracker.g.pidYaw2Srv.reset_filter() and tracker.g.pidPitch2Srv.reset_filter() added to enter() it passed every run, and the merge-base code still fails the test. Continuous-rotation servos get the same kick directly. The test comment saying AUTO holds the bearing (antennatracker.py:129) could also be reworded: AUTO isn't running the controller there, so the PID state isn't continuous across the switch.

Minor: the title still says "wait for first command". The new code holds the current attitude, so a retitle would help.