Plane: use dead zone in stick mixing

[tridge]

prevent small stick offsets causing navigation errors

[IamPete1]

Good idea. Will need testing in every mode, I would be un-surprised to find other cases were were not updating them.

We might need to cache old values for MAVLink to report?

[tridge]

We might need to cache old values for MAVLink to report?

I think the mavlink output will always happen after the mode up, but if mavlink moved to its own thread we could be in trouble.
I'm more concerned about possible access to these variables in scripting and things like FrSky telem. I haven't traced through any possible issues there yet

[tridge]

another potential problem is code which reads nav_roll_cd to determine what to do, see for example Plane::servos_auto_trim(). That case is OK I think as it is called late, but there may be others
@IamPete1 maybe we need a set_nav_roll() and set_nav_pitch() call, which set a boolean to say they have been updated, and if false when we output then zero?

[tridge]

check rudder for dz usage