Performance improvements for Graham's scan

[b4l]

• I agree to follow the project's code of conduct.
• I added an entry to CHANGES.md if knowledge of this change could be valuable to users.

---

[urschrei]

Similar to my comment in #1057, could you say a little about how this improvement works?

[b4l]

Theoretically, it should be an improvement as it may use fewer orientation invocations. The benches show mixed results though, with gains in the synthetic one while regressing the real-world ones, hence a draft. It may need a certain amount of points to become beneficial.

[frewsxcv]

Thanks for opening this @b4l. My opinion is that we'll only consider a change like this if we can consistently see it being a performance win in the benches. So considering that, and that momentum has slowed on this pull request, I'm going to close it. If you feel like picking this back up, feel free to reopen.

[b4l]

@frewsxcv, that is fine. I just left it like this in case someone else follows this path. Should the branch be deleted or kept?

[frewsxcv]

I'll delete it but we can easily restore it if someone else picks it up.