WIP - Fix Porter-Duff Compositors

[JimBobSquarePants]

Prerequisites

• I have written a descriptive pull-request title
• I have verified that there are no overlapping pull-requests open
• I have verified that I am following matches the existing coding patterns and practice as demonstrated in the repository. These follow strict Stylecop rules 👮.
• I have provided test coverage for my change (where applicable)

Description

By my understanding our Porter-Duff equations are based upon the work here.

http://ssp.impulsetrain.com/porterduff.html

I've opened this PR to get work started on a fix.

• Fixed Spelling of Subtract
• Added input test images to match samples on site.
• Added basic test to create output - (no comparison so far)
• Fixed algorithm.

This is what we are looking for. Only a few of our output types are correct which leads me to think we've altered something fundamental when translating the formula. I think the input table values are correct.

I'm gonna need help on this as I can't make head-nor-tail of our interpretation of the formula. I'm a little concerned we'll need to create normalized vector methods for our IPixel implementations for it to work also.

[codecov]

Codecov Report

Merging #493 into master will increase coverage by 0.03%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #493      +/-   ##
==========================================
+ Coverage   86.86%   86.89%   +0.03%     
==========================================
  Files         852      853       +1     
  Lines       36071    36095      +24     
  Branches     2654     2654              
==========================================
+ Hits        31332    31364      +32     
+ Misses       3978     3972       -6     
+ Partials      761      759       -2

Impacted Files	Coverage Δ
.../PixelFormats/PixelBlenders/PorterDuffFunctions.cs	100% <ø> (ø)	⬆️
tests/ImageSharp.Tests/Drawing/DrawImageTest.cs	100% <ø> (ø)	⬆️
tests/ImageSharp.Tests/TestImages.cs	100% <ø> (ø)	⬆️
...ts/PixelBlenders/DefaultPixelBlenders.Generated.cs	96.42% <100%> (ø)	⬆️
...ats/PixelBlenders/PorterDuffFunctions.Generated.cs	78.18% <100%> (ø)	⬆️
...Tests/PixelFormats/PixelOperationsTests.Blender.cs	100% <100%> (ø)	⬆️
...lFormats/PixelBlenders/PorterDuffFunctionsTests.cs	100% <100%> (ø)	⬆️
...s/PixelBlenders/PorterDuffFunctionsTests_TPixel.cs	100% <100%> (ø)	⬆️
...xelFormats/PixelOperations{TPixel}.PixelBenders.cs	95.45% <100%> (ø)	⬆️
...Formats/PixelBlenders/PorterDuffCompositorTests.cs	100% <100%> (ø)
... and 5 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d25c9dc...2b36c6a. Read the comment docs.

[dlemstra]

@JimBobSquarePants Can we merge in the grammatical errors? To make sure this doesn't get to big? Maybe a new PR and rebase this one?

[JimBobSquarePants]

Perhaps, I could be wrong though and the fix could be super simple.

[antonfirsov]

@JimBobSquarePants great analysis of the problem! To summarize things further:

This is how our output looks like with following the original order in the article:

The list of the failing Blenders:

• ATop
• In
• Out
• Dest Atop
• Dest In
• Dest Out

[JimBobSquarePants]

Go team! 🥇