doubled injections possible on negative fueling transients

[mck1117]

crank to run transition

Note that the duration of these two intervals is approximately the same:

What's happening is that the first small injection is scheduled, then the injection phase is updated, then at the next tooth its scheduled AGAIN. Since we target the end-of-injection timing, the first injection got scheduled expecting the duration of the longer cranking pulse, not the shorter running pulse.

[mck1117]

TODO: verify that this still exists in the same form