Update volume visualization to handle floating point data

Summary:
Solves T30364. I replaced the manual histogram calculation with the mitk::HistogramGenerator and adapted some places to handle the potential floating point output.
This introduced some magic numbers, some of which I am unsure about. To get the range slider above the histogram widgets to correctly adapt its step size, I introduced a fixed number of positions (1000), based on which it calculates the step size from the current histogram. This feels alright to me, since 1000 points in the slider seem more than enough.
Regarding the number of bins, I am undecided. At the moment, it is set fixed to 256. For diverse data, this should be enough. For data with very few unique values (e.g. binary images with 0 / 1), this is way too much, though. It still works, which is why I opened the diff anyway, but I am open for suggestions how we could better handle this.

Test Plan: Try the Volume Visualization. With regular data, behavior should be like before. With floating point data, especially between 0 and 1, it should now also work.

Reviewers: O3 MITK Reviewer Group II, kislinsk

Reviewed By: O3 MITK Reviewer Group II, kislinsk

Subscribers: floca, kislinsk

Differential Revision: https://phabricator.mitk.org/D907