Restructure astroquery

[cdeil]

I propose to restructure the astroquery files like this:

• astroquery/service/service.py is renamed to astroquery/service/core.py
• Tests go in astroquery/service/tests/, not all in astroquery/tests.
• Imports are all relative, even from tests (like in astropy)
• Names are not (partially) repeated, e.g. astroquery/simbad/sim_parameters.py is renamed astroquery/simbad/parameters.py.

@keflavich @astrofrog What do you think?

I can do it now, but I don't know how rebasing will go for existing PRs. Do you want to merge e.g. #68 and #70 now and then we just keep working on them in new PRs after the restructure?

[keflavich]

I agree with all points laid out here. Perhaps merging those big PRs is not a bad idea; we'll just have to open new issues soon but if you're going to go ahead with the refactor today, I'd be inclined to let you do it against a cleanly merged master.

[cdeil]

@keflavich @astrofrog Someone please review.

[keflavich]

Any idea why this file is all green? I thought we had it already.

[cdeil]

But I moved it from astroquery/tests/test_fermi.py to astroquery/fermi/tests/test_fermi.py.
Apparently I changed the content in the same commit when I added the license line and now github shows this as a deletion and addition instead of a move.
Not nice, but I suggest to leave it like that ... I'll be more careful not to move and change content in the same commit in the future.

[keflavich]

I didn't see any serious issues, but with so many small changes I'm not sure I would have caught any. Note that all the travis tests are failing, though, even some that were passing before, e.g. https://travis-ci.org/astropy/astroquery/jobs/6338243. I'm seeing gobs of errors with irsadust, which I suspect are caused by the refactor in some way.

The only non-irsadust issue is this:
E AttributeError: 'MaskedArray' object has no attribute 'to_table' in SimbadResult... investigating.

[keflavich]

It might be best to find a different name for this module - maybe leave it as sim_votable.py since we use astropy.io.votable as votable - this may lead to confusion

[cdeil]

Done.

[cdeil]

I haven't figured out the causes for the test failures until now ... if you have a tip that would be great.
I don't have much time, but I'll try some more.

[cdeil]

I fixed the irsa_dust test errors. I forgot to rename tests/t to tests/data in one place.

The remaining simbad test error is not new, I believe. This test just wasn't executed before, because the test function name didn't start with test. Here is the last green travis test run and you can see that the simbad tests aren't run.

My suggestion would be to merge this now and I'll open a new issue for that problem?

[cdeil]

The Python 3 tests on travis-ci also fail due to #80 .
@keflavich Since travis-ci tests for master are red at the moment anyways, are you OK with merging this now and then fixing the issues one by one in the next days?

[keflavich]

Yes, that sounds reasonable. Merge underway.