Implementing Midtones normalisation

[Lasloruhberg]

This PR implements the fitsbolt midtones normalisation option into the ui and main program.
This allows for an overall higher versatility of this program and might allow for more stable cutout appreance over a wide variety of objects.
This PR boils down to 3 changes:

1. Implementing the Midtones option into the validate config as a viable parameter

2. Updated the UI to allow midtones in the dropdown and use the a parameter for the desired scale + percentile

3. Added documentation

[gomezzz]

note to self: tests are only failing due to lack of permissions because PR comes from a fork

[Lasloruhberg]

Unsure how this affects the failing tests of the catalogue_preprocessor. Tests locally pass. I assume this might be a permission error?
This seems to be a pandas version issue and the handeling of object types wrt strings:

This test failed with pandas v 3.0.0 (released yesterday)
Requirement already satisfied: pandas>=1.3 in /home/runner/micromamba/envs/cutana/lib/python3.11/site-packages (from cutana==0.2.1) (3.0.0)

The test on main passed with pandas v 2.3.3
Requirement already satisfied: pandas>=1.3 in /home/runner/micromamba/envs/cutana/lib/python3.11/site-packages (from cutana==0.2.1) (2.3.3)