Draft: feat: implement a safe wrapper for boolops

[RobWalt]

• [x] 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.

I've been playing around with an idea I had recently to fix up the BoolOps with a wrapper. The issue with the current implementation of BoolOps is some inconsistency which is triggered if one point of the left hand side geometry lies on top of a line of the right hand side geometry. To prevent this I just wrote a wrapper which checks if that's the case and if so, moves the point a tiny bit to a place where it's "safer", i.e. into a region where no line of the left hand side geometry is near.

The PR currently includes a lot of tests which I accumulated from all the issues that are open regarding this matter. They all seem to work. However, be aware:

• Because of how it's implemented, this wrapper might produce some undesired performance overheads
• The algo doesn't really work well for huge or tiny scales. It is recommended to scale your geometries so that their coordinates are floats in the range of 0.001..=1000.0 and then scale back after the operation was applied
• I noticed that if the two geomtries have differing magnitudes of scale (as in the test example of 1064) then the algo can fail to produce an intersection (in this example the intersection is empty even though that's technically not true ... but it's like asking for the intersection of a dart head on a darts board)

I'll clean this up a bit further in the coming days:

• [x] docs
• [x] include reasons above in the docs
• [x] update doc comment of trait
• [ ] additional test
• [ ] post some pics

[urschrei]

@rmanoka almost certainly of interest to you

[RobWalt]

I'll add some comments tomorrow so I would highly recommend waiting until then before trying to understand everything just from the code 😅

[RobWalt]

I've tested it and that still doesn't seem to work. Here's a simple app to test it out

https://github.com/RobWalt/is-geo-boolops-still-broken

[frewsxcv]

@RobWalt Did you mean to close this?

[RobWalt]

Yes, both have flaws and special cases that still fail. It's no use and we probably need another approach. I'm working on something though