Consistent aliases

[kvark]

Current approach has RowMatrix aliasing with ColMatrix, yet there is Vector without Point. We should be consistent about having aliases.

• either have them, so add Point2/3 types
• or disallow them completely, so make a single Matrix type

I'm in favour of no aliases rule. It's not clear what to do with EulerAngles in this case. We can either turn it into a tuple or remove the alias entirely.

[Ralith]

To me, it seems like the primary use case of this library is explicit use in public math-library-agnostic interfaces, which requires directly depending upon it. In that circumstance, newtypes make a lot more sense and don't suffer the awkward tradeoff presented here.

[torkleyy]

@Ralith I think the intention is to only depend on them in libraries, not in actual applications. However, I see that newtypes are clearer here. I'm pretty okay with the solution presented in #3