Add math module.

[tychedelia]

Wraps glam in Python, allowing users to call most processing methods as either: foo(1.0, 3.0, 4.0) or foo(vec2(1.0, 3.0, 4.0)).

Allowing both is helpful -- users can use the vec types to do linalg stuff and avoid having to pass things as v.x, v.y, v.z, etc.

Converts most internal processing_render APIs to use glam rather than explicit f32 args. We should have done this from the start and avoids unnecessary conversions.

We don't currently have matrices here, but they can be added later.