[PATCH v2] wined3d: Avoid using float literals for MAX_FLOAT in GLSL when possible.

[Henri Verbeet]

I don't think special casing specific values adds much, but by itself,
using ARB_shader_bit_encoding when available wouldn't be a bad thing.
Note that we do something very similar for immediate values in
shader_glsl_get_register_name(). The specific issue there was related
to specifying floating-point specials like +/-INF or NaN.