Tightened up the code related to capture suspending on bad guide deviations

This is an attempt to address https://bugs.kde.org/show_bug.cgi?id=430365

I was not able to reproduce the issue, however I did look through the code and tightened up a couple things that will hopefully improve the situation.

• The previous code suspended capture if there were 2 or more guide iterations that exceeded the stated guide drift value, even if they were separated by hours (in a single capture sequence). This now requires 3 CONSECUTIVE guide iterations to exceed the value in order to suspend capture
• The current code exempted the states GUIDE_DITHERING and GUIDE_MANUAL_DITHERING from broadcasting its drift to capture, but didn't exempt GUIDE_DITHERING_ERROR, GUIDE_DITHERING_SUCCESS, and GUIDE_DITHERING_SETTLE. It's possible that some drift computed in one of the latter states found its way to capture, and caused capture to suspend.