Metric cellcharter purity loukue

[LouisK92]

Okay so this PR unfortunately doesn't completely follow the template restrictions yet. The reason is that there needs to be done some checks to set one part of the script.

So for the metric calculation we first need to compute connected components based on the spatial domains and connectivities between cells.
In the script I added these steps. The problem is that it's unclear how well the current component computation generalizes over different datasets/tissues/domains. The script is now organised in such a way that it optionally generates a plot where you can see the domains, I would check this for a few datasets and connectivity calculation settings, and then decide on the connectivity computation based on these outputs. (I attached 3 plots of examples that were run with connectivity="neighs")

Besides that, note that all other parameters that are set in the script are fairly well double checked to generalise well (clarified this with Marco Varrone), also implemented some additional heuristic for the boundary computation of the components.

So there are three things to adjust in the script:

1. Decide on the connectivity calculation
2. (optional) throw out the debugging functionalities, i.e. mainly the plot function
3. adjust the input arguments

Additional possibilities

• Maybe it would be useful to define some test on the connected components and return 0/NaN for the metric if the components don't pass the tests.
• provide components as input instead of calculating them within the metric script.

I think with the current form of the script it's not too difficult to take it on from here and finalise it after having more datasets/outputs for generated spatial domains.

some plots of examples runs

[LouisK92]

ah, refering to #74

[LouisK92]

a test script is here: scratch/SpaceHack2/test_scripts/test_metric_cellcharter_purity.sh