MulticopterLandDetector: Make land detection time configurable

[MaEtUgR]

Describe problem solved by this pull request

1. Land detector trigger times are hardcoded values:
   The tree stages used arbitrary 350, 250 300ms totally 900ms
2. It's hard and error-prone to debug the land detector when you have to calculate each condition manually at every point in time.
3. An efficiently designed multirotor commanded to hold constant altitude but moving fast horizontally can during hard braking maneuvers hit the limit of low thrust because of the additional lift generated by the horizontal movement airflow around the (rather large) propellers. In that case it often has worse altitude tracking and the position controller demands additional downwards velocity to reach the right altitude again. This is with the current land detector implementation intentionally interpreted as the intent to descend with the vehicle. It's done that way to implicitly allow to land by moving the position setpoint into the ground without setting a vertical velocity setpoint but we don't use this method of landing anymore for quite a while now.

Describe your solution

1. Make the total land detection time configurable and each stage a fraction of it.
   By default 1 second -> 333ms per stage.
2. I ported @RomanBapst 's commit to log the necessary flags to easily understand what's going from a log.
3. I switch the descend detection to look at the trajectory's velocity setpoint to eliminate the problem. During a hard brake with too much lift all the flags about low thrust, movement might go to true but if there's no intent to descend landing will not be detected. This has the disadvantage that landing with an altitude setpoint only that is moving into the ground is not possible anymore. Wherever that might occur. We could in principle add an extra condition for that case but I didn't do that here.

Test data / coverage

1. SITL tested with output of the concrete us value and testing with 1 second and 10 second trigger time.
2. Log shows all the flags correctly, we use that for quite some time now.
3. I SITL tested and any normal, expected way of landing uses a downwards velocity setpoint and works exactly like before.

[sfuhrer]

Cool stuff, I agree that we need the land duration setting exposed, as this can be an easy to use setting to reduce the amount of false positives for land detection on low throttle systems.

[MaEtUgR]

Thanks for all the good reviews! I tested the configurable time and in decend changes also on a real vehicle and followed up on the review comments.

Because #17403 was quickly merged in between I had to slightly change the distance sensor logic to apply the parameter changes in a consistant way:
3f2c631#diff-13eefebea7bf72c2fa6a4192610df3e813b41098dff9b4b9c55c7cdb77ae0554R105-R110

It would really be better to keep these things separate since they are in principle used by the multicopter land detector only.

Ready for final review?

[MaEtUgR]

How was I able to commit with a style check failing missing newline 🙈
https://github.com/PX4/PX4-Autopilot/compare/90139c680c2bad342e539678134a4fb8b492796c..5e9b873481d2ea88462bf3ac81c27f05db15aca1

[MaEtUgR]

Merging since it was reviewed before adding the empty line for style.