GSVertexTrace::FindMinMax improvements

[TellowKrinkle]

when you spend a bunch of time optimizing a function because you think it accounts for 30% of runtime but that was actually due to the compiler being stupid and not issues with the algorithm

This PR contains the following changes:

• Prevents clang from optimizing out our denormal-removal shuffles (10x faster than before for people who compile with clang!)
• Run divides on four elements at a time instead of two elements and two useless numbers
• Run per-vertex instead of per-primitive for all things that need to apply to all vertices of a primitive anyways (prevents double-checking vertices in triangle strips, and also reduces pointer chases in the main loop) This broke stuff
• Remove inaccurate stq
  • With the above division improvements, on processors with partially-pipelined division (Ivy Bridge and later, Bulldozer and later), accurate stq is actually faster (according to both IACA of inaccurate vs accurate and LLVM MCA). On older CPUs expect performance to be about 2/3 of the old algorithm before taking into account improvements from not double-checking vertices.
  • There seems to be a check of accurate_stq when the OGL backend is deciding whether to use geometry shaders to process sprites, does anyone know if that's important or just something someone threw in there that's okay to remove? Reason found, added as comment

In the end, ignoring clang issues, GSVertexTrace::FindMinMax goes from taking about 3% of MTGS thread runtime to 1.5% on my computer. (Most of the time was spent doing OpenGL things so if you have a more efficient OpenGL driver it might make more of a difference for you)

[TellowKrinkle]

This brings the slowdowns on Xenosaga Episode 1 during the third cut scene from ~40-45 fps to ~50-55 fps when using clang11.

Curious, after this patch, are clang builds slower than gcc builds here?

[frymezim]

@iMineLink or @gregory38 might be able to help review this pull request.

[lightningterror]

I'd add a prefix to the newest commit gsdx-ogl:.

[lightningterror]

Would be nice to rename GSdx to GS in commit names as GS was merged.

[refractionpcsx2]

Been messing with this (while reimplementing the CLAMP optimisation), it looks good to me, everything seems to work good :)

Edit: There is a bug, will tell you about it on discord

Edit2: Nope, it was VS screwing me over, the PR is fine.