Support allenes in canonicalizing double bonds, fixes #7044

[pechersky]

Reference Issue

Fixes #7044

What does this implement/fix? Explain your changes.

During canonicalization of double bonds, neighbor bonds have to be found to correctly label the stereochem. During #6643, the allene case was not supported. This case comes up during a Tautomer Canonicalization of conjugated nitro or oximes. So I added tests around those (ones that came up in the issues and real-life use cases).

Any other comments?

I tried to make changes in the actual Tautomer Canonicalization to not copy stereo atoms in allene situations, but it seemed more direct to avoid a nullptr situation and hope that that is doing the right thing. The new tests should shore up that the hope is true, and the rest of tests passing means that it is a targeted fix.

[greglandrum]

@pechersky Thanks for the contribution!

I just made a PR against your branch with a suggested alternate formulation of the fix; take a look and see what you think.

[pechersky]

Thanks @greglandrum! I appreciate the direct fix. I had kept my fix localized to allenes because I was afraid of a similar issue popping up in other molecules which would have helped expose some other issue. If you're feeling good about exiting early in any such situation, I'm good with that too.

[pechersky]

I've been trying to test locally, but have been hitting some cmake troubles (since the last time I had it working):

[...]
-- Found RingDecomposerLib source in /home/yakov/orion/rdkit/External/RingFamilies/RingDecomposerLib/src/RingDecomposerLib
CMake Error at Code/GraphMol/SmilesParse/CMakeLists.txt:51 (CONFIGURE_FILE):
  Operation not permitted


CMake Error at Code/GraphMol/SmilesParse/CMakeLists.txt:53 (CONFIGURE_FILE):
  Operation not permitted


== Updating Filters.cpp from pains file
[...]

So, excuse my clumsy CI-testing in the meantime please =)

[pechersky]

@greglandrum Would be happy to have a re-review. The implementation comes from your PR.