Implement some of the straightforward data types and functions

[mwcraig]

This is not ready to merge yet -- doing the pull requests to get automatic builds on Travis.

• Remove the xfails on tests that need a fix for Make initialization of results in arithmetic and unit conversion to allow more flexibility in subclasses astropy#2301

Closes #14 with 7ff79bc

[mwcraig]

Just remembered, this only works with an astropy fixed with something like astropy/astropy#2301 so all tests will fail for now.

[mwcraig]

Put in an xfail on the tests that will break until the underlying astropy problem is addressed.

[crawfordsm]

Let me know if you want me to merge this now or once nddata is updated. Also are you able to merge things?

[mwcraig]

This one can sit for a while--I'm planning to put the API changes I'm working on for next week into this too.

I can merge -- the other one I asked you to do just to make sure you saw the changes since they affected how tests are written.

I'd like to wait for the nddata changes to be merged in to astropy, as long as it doesn't take too long.

[crawfordsm]

Good to see--I'm also wondering if at the same time it would be worthwhile to put in a pull request/issue to nddata about being able to use quantity objects for the array as well? It probably would be good and related to how the units are being treated.

[mwcraig]

I'm guessing the nddata/unit discussion would be quite a bit broader (see astropy/astropy#1314 ) but I'm thinking it is time for NDData to see some focused attention. In addition to the we're having, Stuart Mumford posted about a separate but related issue on astropy-dev and there is astropy/astropy#2170 which is essentially another initialization issue.

[mwcraig]

Travis builds are succeeding now, so I'm merging this in.