Include or not adjacent cells when using extract_points method

[mglesser]

Overview

An input parameter ("adjacent_cells") is added to the "extract_points" method. For backward compatibility it is set to "True" by default. In this case, the output includes the cells that contain at least one of the extracted points. When it set to False, the output includes the cells that contain exclusively points from the extracted points list.
Please note that when using quadratic mesh, the list of nodes shall include the corner nodes AND the middle nodes for the adjacent_cells = False case.

A test function is provided to check the behavior of the "extract_points" method on a simple mesh.

resolves #953

[codecov]

Codecov Report

Merging #960 into master will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #960      +/-   ##
==========================================
+ Coverage   86.95%   86.98%   +0.02%     
==========================================
  Files          36       36              
  Lines        9047     9065      +18     
==========================================
+ Hits         7867     7885      +18     
  Misses       1180     1180              

[akaszynski]

Thanks for adding this and the test as well.

[math-artist]

I have tried using this new function and it keeps missing one point. I can't figure out where this is happening.

len(ind) at the beginning is always one more than _get_output(extract_sel).n_points.

This is causing me issues for adding filtered scalars. I have checked numpy code, numpy_to_idarr and everything is ok. This data loss is probably happenning during the extract:

        selection = vtk.vtkSelection()
        selection.AddNode(selectionNode)

        # extract
        extract_sel = vtk.vtkExtractSelection()
        extract_sel.SetInputData(0, dataset)
        extract_sel.SetInputData(1, selection)
        extract_sel.Update()
        return _get_output(extract_sel)