Fix #5895: Properly clip MOVETO commands

[mdboom]

I haven't run the tests on this yet... we'll see what Travis CI thinks.

[WeatherGod]

Travis failed with lots of small differences.

[mdboom]

This is passing Travis now and ready for final review.

[tacaswell]

what does this line do? My attempts to understand this by writing code have not gone well

// switch_statement1.cpp
#include <iostream>

enum test_e
  {
    A, B, C
  };



void helper(test_e inp)
{
  switch (inp)
  {
  case (A | B):
    std::cout<< "One"<< std::endl;
    break;
  case C:
    std::cout<< "Three"<< std::endl;
    break;

  default:
    std::cout<< "NONE?!"<< std::endl;
  }
}

int main() {
  helper(A);
  helper(B);
  helper(C);

}

gives

22:07 $ ./a.out
NONE?!
One
Three

[tacaswell]

with some digging I now see that this filtering with two things that just look like they are members of the same enum set.

[mdboom]

The value agg::path_cmd_end_poly | agg::path_flags_close is how Agg indicates the end of a path that is also closed. Why that's not a single enum value, I don't know, but that's how it's done in Agg so we must follow. If it makes the code clearer I could create a new alias:

end_and_close_poly = agg::path_cmd_end_poly | agg::path_flags_close

[tacaswell]

Nah, I am just on a wading-into-c-code-I-do-not-understand kick the last few days.

[tacaswell]

We need some way to mark that this move is 'synthetic' so that the logic an line 366 knows not to re-set the m_initX / m_initY values. I think that would allow us to clip polygons and fix #5948 more gracefully.

[tacaswell]

back ported as 512c8d0