Deprecate and remove PlotterITK

[banesullivan]

I'm pretty sure no one is actually using this, and it wasn't even registered as a jupyter_backend. Maintaining this is more of a burden than a benefit, and there are much better alternatives to this API... like using itkwidgets directly which supports PyVista data types!

This is in support of #3690

We're going to deprecate and remove this immediately without a transition plan or warning because it is not a core feature and not a proper "jupyter backend"

[banesullivan]

Further justification: itkwidgets is broken on import and causing CI failures:

==================================== ERRORS ====================================
_____________ ERROR collecting tests/jupyter/test_itk_plotting.py ______________
tests/jupyter/test_itk_plotting.py:11: in <module>
    import itkwidgets
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itkwidgets/__init__.py:13: in <module>
    from .widget_viewer import Viewer, view
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itkwidgets/widget_viewer.py:14: in <module>
    import itk
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itk/__init__.py:28: in <module>
    from itk.support.extras import *
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itk/support/extras.py:35: in <module>
    import itk.support.types as itkt
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itk/support/types.py:172: in <module>
    ) = itkCType.initialize_c_types_once()
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/itk/support/types.py:137: in initialize_c_types_once
    _B: "itkCType" = itkCType("bool", "B", np.bool)
/usr/share/miniconda/envs/pyvista-vtk/lib/python3.8/site-packages/numpy/__init__.py:284: in __getattr__
    raise AttributeError("module {!r} has no attribute "
E   AttributeError: module 'numpy' has no attribute 'bool'
=========================== short test summary info ============================
ERROR tests/jupyter/test_itk_plotting.py - AttributeError: module 'numpy' has no attribute 'bool'
!!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!

Looks like it is incompatoble with numpy 1.24 which gave ample warning about the built-in type aliases going away

With this in mind, I'm treating this as a HOTFIX and will merge expeditiously

[banesullivan]

CI failure due to needing #3736

[codecov]

Codecov Report

Merging #3737 (39e5cdb) into main (4bc2064) will decrease coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #3737      +/-   ##
==========================================
- Coverage   94.04%   94.02%   -0.02%     
==========================================
  Files          83       82       -1     
  Lines       18643    18545      -98     
==========================================
- Hits        17532    17437      -95     
+ Misses       1111     1108       -3     

[adeak]

I'm surprised by this breakage since in #3659 (comment) I could get rid of these by upgrading all dependencies...

[adeak]

I'm also surprised we don't pin the basic dependencies. If numpy were bumped by dependabot we would've seen this coming.

[banesullivan]

I'm also surprised we don't pin the basic dependencies.

Oh huh. We don't pin numpy, do we. At first, I thought this was just that we don't pin the conda environment but yeah we don't pin the core dependencies. We may want to look into that seperately

[akaszynski]

Agree with removing this. This has been broken ever since JupyterLab 3.0, and we have way too many juptyer_backends and should start to converge on a consistent backend soon.

Recommending immediate merge to fix the build.