coverage: 58.939% (-0.07%) from 59.007%
when pulling e30ea42 on eyal0:path_manager2
into 6c8e590 on pcb2gcode:master.

Progress on #733.

I can confirm that eulerian_paths are the problem. Disabling the feature is no good because it makes the code way too slow. I have fixed up the problem a little bit already.

Here's what the output looks like on master branch for one example.

Here's what it looks like on tag 928a7809:

Better. And here's what it looks like if I disable eulerian paths on line 519 of surface_vectorial.cpp:

It ran a little slower but it is perfect. It will also mill slower due to not joining paths.

I will write try to whittle down that run of the eulerian algorithm to figure out what's going wrong.

Oh no the test case is 11000+ linestrings. It's so big that I can't compile it. I have to read it from a file dynamically for now. :-(

Done in #740