Simplify transforms addition.

[anntzer]

• Suggest tr1 + tr2 instead of composite_transform_factory(tr1, tr2).
• No need to implement Transform.__radd__ -- we can only add
  transforms so __add__ is enough.

PR Summary

PR Checklist

• Has Pytest style unit tests
• Code is Flake 8 compliant
• New features are documented, with examples if plot related
• Documentation is sphinx and numpydoc compliant
• Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
• Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way

[timhoffm]

Just to be sure, this was thought about: We want tr1 + tr2 and not tr1 @ tr2?

Concatenating transforms is a bit matrix multiplication-like. OTOH + is often used as general join operator (strings, lists, ...). Overall, I think @ would be a bit too sophisticated, and + is the plain simple solution to go.

@anntzer may self-merge if you agree.

[dopplershift]

Concatenation is multiplication for affine transforms. But it's completely not for the non-affine case. While the @ is intriguing, I think because it doesn't apply in general we should stick with +. If someone wanted @ as an alternate, I wouldn't protest.

[anntzer]

A problem is that the order of the operands would have to be inverted (A@B typically means "apply B then A" in linear algebra, but that's spelled B+A right now).
In any case tr1 + tr2 is very widely used in the library right now and this PR is not about changing that, it's just about using it even more.