maint: unify centroid computation

[AhmetNSimsek]

This PR fixes #613 by using the robust centroid computation used in Region.spatial_props(). Now all centroid computation is done by finding the components of an image with skimage.measure.label.() and then finding the center of the non-zero values in the voxel space.

Also:

• volume.ComponentSpatialProperties is introduced in place of region.SpatialPropCmpt.
• region.SpatialProp is removed since it is not used and was only introduced for typing.
• added tests
• Allow user to not split to components

Not backward compatible:
Instead of using SpatialProps.components now the result of Volume.compute_spatial_props() is a list of ComponentSpatialProperties per component. If the image is not split to components then it will return a single ComponentSpatialProperties.

Note: this PR does not address #157. The root cause of this issue is the concaved geometry of the regions.

[xgui3783]

LGTM

[codecov]

Codecov Report

Attention: Patch coverage is 29.16667% with 34 lines in your changes missing coverage. Please review.

Project coverage is 46.00%. Comparing base (8da694f) to head (dc63457).
Report is 162 commits behind head on main.

Files with missing lines	Patch %	Lines
siibra/volumes/volume.py	34.61%	17 Missing ⚠️
siibra/volumes/parcellationmap.py	9.09%	10 Missing ⚠️
siibra/core/region.py	16.66%	5 Missing ⚠️
siibra/locations/point.py	33.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #610      +/-   ##
==========================================
- Coverage   51.64%   46.00%   -5.64%     
==========================================
  Files          74       75       +1     
  Lines        7012     7232     +220     
==========================================
- Hits         3621     3327     -294     
- Misses       3391     3905     +514     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.