Fix principal direction extent

[eleftherioszisis]

Fixes #1007

[eleftherioszisis]

@lidakanari , could you please check that with these changes you get the expected results?

[codecov-commenter]

Codecov Report

Merging #1008 (c3b3b1d) into master (fdf9f78) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master     #1008   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           36        36           
  Lines         2355      2350    -5     
=========================================
- Hits          2355      2350    -5     

[lidakanari]

The function seems to return correct values for the tested cases. However, the signature of what it returns changes with respect to the previous one, as it now returns a vector of the same dimension as the input data. i.e. 2d point cloud returned 3d vector before, now it returns 2d.

[eleftherioszisis]

The function seems to return correct values for the tested cases. However, the signature of what it returns changes with respect to the previous one, as it now returns a vector of the same dimension as the input data. i.e. 2d point cloud returned 3d vector before, now it returns 2d.

Indeed, the dimension of the extents depends on the dimension of the points. Therefore, for 2D points it would return a 2D vector of extents, and for 3D a 3D respectively. For NeuroM purposes, only 3D points are used, so this is not really an issue.

[lidakanari]

Ok, that should be good. Also something to keep in mind, currently the extents are not sorted (i.e. largest to smallest). If this is by choice, it would be good to mention in documentation because it might be confusing.

[eleftherioszisis]

Ok, that should be good. Also something to keep in mind, currently the extents are not sorted (i.e. largest to smallest). If this is by choice, it would be good to mention in documentation because it might be confusing.

That's true indeed. There was no requirement that they are sorted from max to min, which imho renders the direction argument a bit useless in the respective feature.

Should I order them?

[lidakanari]

I think you can implement a "major_extend" feature or order them (but that would be a breaking change).