Position Control Refactoring Part 2

[MaEtUgR]

Describe problem solved by this pull request
This is part 2 of the refactoring from the monster pr #12072 and should be easy to understand and review since the diff is manageable.

Describe your solution

• Changing the acceleration and jerk in the vehicle_local_position_setpoint message to arrays because they most of the time are loaded into Vector3fs immediately anyways and it doesn't make sense to write x, y, z out every time. Same goes for position and velocity but the interface to flight review should be switched in one go and not message by message so I'll combine that with RC: Have all sticks scaled from -1 to 1 in the normalized representation #9331
• The WeatherVane library needs just the last column of the attitude rotation matrix so it's not necessary to keep the whole attitude setpoint and pass the quaternion. I refactored it to take just that last column.
• Since the attitude setpoint is not needed as a position control state it can be filled on the stack and published.

Test data / coverage
SITL tested except for the WeatherVane library but double-checked it gets the exact same data.

Additional context
Part 1 that goes before this: #13247
Part 3 that comes after this: #13262

[bresch]

I didn't find anything to criticize :D

[bresch]

@MaEtUgR Can't wait for part 3 🤓

[MaEtUgR]

@bresch Thank you! 👍
Here is part 3: #13262