vectorize bounding box query

[giovp]

Trying to vectorize the tiling in the dataloader PR, realized some improvements should be added separately.

With this PR, I enable the vectorization of bounding_box_query for all elements.

This means that it is now possible to pass an array of bounding boxes (and not just one).

TODO:

• tests for multiple bounding box queries for raster data
• tests for multiple bounding box queries for shapes
• tests for multiple bounding box queries for points

@LucaMarconato @melonora do you have any suggestion of when this could be used to replace current implementations across the projects? A clear use case (and the motivation for this contribution) is the dataloader, see #687 for the speedup, but I wonder if there are other places where this is useful.

This also is the groundwork for eventual update to the dataloader, being able to return batches of all elements.

considering that this PR is already getting too large, I would postpone the vectorization of polygon query.

[codecov]

Codecov Report

Attention: Patch coverage is 91.75258% with 16 lines in your changes missing coverage. Please review.

Project coverage is 91.76%. Comparing base (774b492) to head (b8b6331).
Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_core/query/_utils.py	82.05%	14 Missing ⚠️
src/spatialdata/_core/query/spatial_query.py	98.14%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #699      +/-   ##
==========================================
- Coverage   91.83%   91.76%   -0.07%     
==========================================
  Files          44       45       +1     
  Lines        6781     6887     +106     
==========================================
+ Hits         6227     6320      +93     
- Misses        554      567      +13     

Files with missing lines	Coverage Δ
src/spatialdata/_docs.py	100.00% <100.00%> (ø)
src/spatialdata/_core/query/spatial_query.py	95.21% <98.14%> (+0.47%)	⬆️
src/spatialdata/_core/query/_utils.py	84.37% <82.05%> (-11.28%)	⬇️

... and 1 file with indirect coverage changes

[giovp]

ready for a first pass of review

[giovp]

ready for review

[LucaMarconato]

Thank you @giovp for the PR, I will review now.

A clear use case (and the motivation for this contribution) is the dataloader, see #687 for the speedup, but I wonder if there are other places where this is useful.

I can't think of other places right now, so I think we are good with the current improvement.

considering that this PR is already getting too large, I would postpone the vectorization of polygon query.

I agree.

[LucaMarconato]

I think nb.types.Array[nb.float64, np.float64] may be incorrect and that the correct version is nb.types.Array(nb.float64, 2, 'C'). But I am not sure because pre-commit doesn't complain, so maybe both syntaxes are allowed.

[LucaMarconato]

Mine is wrong. Are you sure about your typing? What looks strange to me is that types like nb.types.Array[nb.float64, nb.int64] would not have a meaning. I tried using just nb.types.Array and pre-commit works, maybe this is the way to go.

[giovp]

I will look into it!

[LucaMarconato]

I removed the dtype, I'd merge.

[LucaMarconato]

@giovp I finished reviewing; I applied some minor changes like simplified the docs and added an extra test.