Ensure all internal float types use double precision

[nmwsharp]

General strategy is to use double types internally everywhere, to avoid subtle situations (like IO) where a loss of precision occurs in user data.

We've been shuffling types internally, so make sure no floats (or glm::vec3) slipped through.

[nmwsharp]

Actually, in the current version of Polyscope, we use glm::vec3 everywhere, which means float precision. Perhaps we should switch to double?

[nmwsharp]

Resolution: we (mostly) use float everywhere internally, since that is all the renderer understands anyway. This could possibly cause confusion due to truncation in some situations.