Suspicious code in hasDiagonalArtifactInner

[bluebear94]

I’m trying to port msdfgen’s error correction algorithm to my pure-Rust implementation fdsm, and I found two potential problems with the hasDiagonalArtifactInner function in msdfgen:

• The linear term passed to solveQuadratic subtracts dA twice. I think dBC - 2.0 * dA is what was meant here, but it would be good to at least leave a comment that subtracting by dA twice is intentional and not a typo.
• When checking xm against the interpolated medians at the local extrema, the em variable is never used.

[Chlumsky]

Thanks for finding this. You are right that em not being used is an error. I have committed a fix. dBC-dA-dA is equivalent to dBC-2*dA, yes. If it helps you in any way I might be able to provide you with the "testbed" I originally developed this with.