* removed bus options that were essentially duplicates, causing hardfaults if attempting to start an already running driver. Changed NULL to nullptr. Removed bus options that were unused
* removed condition from if statement that was unnecessary
* fixed style

- use a linked-list instead of std::vector. Insertion and removal are now
  O(1)
- avoid malloc and use a thread_local instance of TimedWait.
  It gets destroyed when the thread exits, so we have to add protection
  in case a thread exits too quickly. This in turn requires a fix to the
  unit-tests.

Because $sitl_command contains quotes

Previously hrt_absolute_time() was at around 5% of the total CPU usage, now
it's around 0.35%.

less function calls

Not required, since the lock is held during the whole loop iteration.

- the loop is not needed
- we optimize for the fast case and lock only if really needed

This is a small method that is used a lot.

    - mavlink v2.0 in PX4/Firmware (4787cabae6fcf55ef0f8bf19b0009de593c844db): https://github.com/mavlink/c_library_v2/commit/90d9b285e01fe8bfa3b4e8868ca71c5537d43302
    - mavlink v2.0 current upstream: https://github.com/mavlink/c_library_v2/commit/339abb5f4c20865ea67dd417b099e3f54683a2ad
    - Changes: https://github.com/mavlink/c_library_v2/compare/90d9b285e01fe8bfa3b4e8868ca71c5537d43302...339abb5f4c20865ea67dd417b099e3f54683a2ad

 - cleanup airspeed validity check logic

 * this should fix the CI failures where we intermittently saw SIGABRT. (fixes #11144)

 * We have RAM problems and can't afford 4 mavlink instances. This only adds logic the rc.board for the fmu-v4 that only has an effect for SYS_AUTOSTART = 4250 (TealOne airframe).

Fixed ICM20948 magnetometer only option.
Specified bus for internal mpu9250 on Pixhawk 2.1 to avoid premature detection of
other device.

- I enabled it by default mostly for testing purposes
- Disabling it reduces resource usage as most setups don't need it

Fixed an issue with a local variable that should have been a member variable. Increased the P term to reduce rise time. Removed feedforward as it was not neccessary. Removed some custom commands that didn't serve much of a purpose.

* increase pmw3901 publish rate and max_ground_distance
* set work queue to high priority

    - mavlink v2.0 in PX4/Firmware (613a5f5e172de071ebd96b4cb5313316a10f8023): https://github.com/mavlink/c_library_v2/commit/175ab6725cc0d96827d04ffeeaf3a7797a9abdbf
    - mavlink v2.0 current upstream: https://github.com/mavlink/c_library_v2/commit/90d9b285e01fe8bfa3b4e8868ca71c5537d43302
    - Changes: https://github.com/mavlink/c_library_v2/compare/175ab6725cc0d96827d04ffeeaf3a7797a9abdbf...90d9b285e01fe8bfa3b4e8868ca71c5537d43302

This should fix:
- HITL regression (Init MAVLink forever).
- Starting using default arguments.

PositionControl: update comment about the order of thrust and position

mc_pos_control: reset setpoints to NAN if required

MulticopterLandDetector: consider to be landed if vehicle is not armed

mc_pos_control: initialize landing struct with landed

mc-pos-ctrl: adjust thrust-setpoint and yaw during ground-contact/maybe-landed without capturing
that information within vehicle-local-position-setpoint topic because that information
does not belong to user intention

PositionControl: set local position/velocity setpoint to NAN if not used in the control pipeline

mc-pos-ctrl: Fill vehicle_local_position_setpoint_s structure as follow:
The message contains setpoints where each type of setpoint is either the input to the PositionController
or was generated by the PositionController and therefore corresponds to the PositioControl internal states (states that were generated by P-PID).
Example:
If the desired setpoint is position-setpoint, _local_pos_sp will contain
position-, velocity- and thrust-setpoint where the velocity- and thrust-setpoint were generated by the PositionControlller.
If the desired setpoint has a velocity-setpoint only, then _local_pos_sp will contain valid velocity- and thrust-setpoint, but the position-setpoint
will remain NAN. Given that the PositionController cannot generate a position-setpoint, this type of setpoint is always equal to the input to the
PositionController.

mc_pos_control: switch to designated initializer for landed

It's less error prone because it produces an error on every discrepancy.

Fix math::constrain usage

Fix second constraint

 - these only run if code coverage is enabled

Since SBUS does not have CRC, we can use timing information to improve
parsing reliability and reject unexpected bytes.

 - <cfloat> isn't available in the NuttX c++ library

 - <cinttypes> isn't available in the NuttX c++ library