Gaussian-fitting occasionally stops working showing a few different patterns of behaviors.

[KiyofumiTakaba]

Also, the drawn ellipses might take a second to be erased, although the fitting was disabled...

Originally posted by @kferjaoui in #120

As mentioned in the parent issue, sometimes the drawn ellipse appears to remain displayed when the fitting is deactivated. This is still unsolved.
In addition, sometimes the fitting seems to be stuck and the ellipse does not track the beam center. This induces the unreliable beam center definition and failure indexing in post-processing regardless of the diffraction quality.
Currently it is not predictable what causes this and when it happens. Some screencaptures would be helpful to solve this issue, if possible.

[KiyofumiTakaba]

Slightly different but relevant, sometimes the identified beam center jumps just before data-collection.
The following two pictures are the very frames before and after the jumping, on starting the data-collection. This does not seem to occur very often, but to be not negligible.