Improve Continuous scatter-plot tetrahedron classification

[Lgt2x]

Thanks for contributing to TTK!

Before submitting your pull request, please:

• Review our Contributor Guidelines, in particular regarding code formatting (with clang-format) and continuous integration.

• Please provide a quick description of your contributions below:

Following similar work on the VTK-m libary, improve the continuous Scatterplot implementation of Shirley-Tuchman tetrahedron classification. This PR also does variable renaming for maintainability.

For each tetrahedron, instead of 4 barycentric coordinates and 3 segment intersections computations in the worst case, all we do now is 4 single-coordinate cross products. The full explanation is provided in the code comments.

This change allows an estimated 20% speedup on the filter execution, rendering non included.

[Lgt2x]

ping @CharlesGueunet for review :)

[CharlesGueunet]

+1 after small question

[julien-tierny]

Thanks a lot for this PR !
May that be the first of many :)

That's great stuff. I only observed a 9% speedup on my end but that's still good to take!!!

Could you please address the few remarks I left in the code?
Thanks!

[julien-tierny]

alright, it's all good to me :)
@CharlesGueunet anything to add regarding code review?

random question:
the speedup on my desktop -- a Xeon E5-2630 CPU with 12 cores -- is more modest than on my laptop -- an i7-10810U with 6 cores (about 3% VS 9%).
what machine did you do you performance experiments on?

[Lgt2x]

I tested both on Intel i7-7700HQ (8 cores) and 12th Gen Intel i9-12900K (24 cores) with double-templating enabled and got a more or less 20% improvement.

For this benchmark however, I excluded the rendering part of the algorithm which takes the most time out of all steps especially for large images, to get the speedup only on the core classification & density computation.

[CharlesGueunet]

Nothing to add on my side :)
( my small question was addressed directly, and unpublished as I missed one button on github)

[julien-tierny]

alright, let's go for it then :)
thanks a lot!