Fixes on dataloaders

[LucaMarconato]

This PR improves the ImageTilesDataset class in the following ways.

• General minor cleanup, validation of the arguments and improvement of the docstring.
• Now the regions to derive the tiles from can also be labels and multiscale lables; before only polygons, multipolygons and circles were supported. Points are not supported.
• Now the dataset works also with multiscale images, both if rasterize is True or False. The output is always a single scale image.
• The implementation is now simplified thanks to the use of to_circles() API; in particular in _get_tile_coords() there is no need anymore to perform affine matrix multiplications manually.
• I removed the logic around polygon.length. Before if the regions were polygons or multipolygons, the perimeter of these polygon would have been used as the tile width/height. This contrasted with the usage of the diameter for circles. Now any region element is transformed to circles element using to_circles() and the tile sizes are the diameters of these circles.

[codecov]

Codecov Report

Attention: Patch coverage is 93.61702% with 6 lines in your changes are missing coverage. Please review.

Project coverage is 92.63%. Comparing base (421c315) to head (a119850).
Report is 1 commits behind head on main.

❗ Current head a119850 differs from pull request most recent head e7aef86. Consider uploading reports for the commit e7aef86 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #439      +/-   ##
==========================================
+ Coverage   92.58%   92.63%   +0.05%     
==========================================
  Files          43       43              
  Lines        5867     5894      +27     
==========================================
+ Hits         5432     5460      +28     
+ Misses        435      434       -1     

Files	Coverage Δ
src/spatialdata/datasets.py	100.00% <100.00%> (ø)
src/spatialdata/dataloader/datasets.py	90.86% <92.40%> (+2.15%)	⬆️

... and 3 files with indirect coverage changes

[LucaMarconato]

CC @giovp

[LucaMarconato]

Further fixes are introduced with https://github.com/scverse/spatialdata/pull/495/files.