Adjust post-stimulus fixation duration based on expected onsets and durations

[tsalo]

There's some onset drift because of lags (probably exacerbated by running this on my laptop with other processes running), so it would be good to adjust the fixation duration, not just based on the trial duration, but based on the expected onset instead.