[Mesa-dev] Matrix flipper v2

[Kenneth Graunke]

This series does two things: first, it plumbs ShaderCompilerOptions into
do_common_optimization(), since optimization & lowering passes may want
to consult that.

Secondly, it adds an optimization pass that converts (matrix * vector)
operations to (vector * matrixTranspose) for certain built-in matrices
(currently gl_ModelViewProjectionMatrix and gl_TextureMatrix).  This
makes it use DP4s instead of MUL/ADD, which can be more efficient.