Fixed wing auto level bug fix

[breadoven]

Closes #9637 and #9615.

Freezes integrator if last update time interval exceeds 5 x expected interval. Also changes I term factor to further attenuate response (may need tweaking further).

Still needs testing.

[breadoven]

Testing on HITL shows the glitch you get when Auto Level is started is fixed by this PR. Reducing the I term factor related to fw_level_pitch_gain probably also helps. There's some margin for adjustment of fw_level_pitch_gain anyway from the default of 5 up to a max of 20 if a faster response is required.

Could do with some more feedback from actual testing ideally but as it stands this fix seems to work as expected.

[Jetrell]

Could do with some more feedback from actual testing ideally but as it stands this fix seems to work as expected.

I tested this with the default gain.. And switched between different modes after the tune was complete. And the fw_level_pitch_trim value remained the same.
While before this fix, using this plane.. I would see it instantly jump to values like -8 and return to 2 again in the tuning process. Now it holds within a small margin.
So it appears to have fixed the problem.

[breadoven]

@Jetrell This fixes the starting glitch, it's more whether the PID I term factor change works for the available setting range. Doesn't seem to cause problems using the default setting in your case so may as well merge this and hope the setting range is enough to avoid problems with more twitchy craft.