Added general API tests

[astrofrog]

Still trying to figure out how to turn this into a markdown table...

Table of doom:

Test	PolygonPixel	EllipsePixel	RectanglePixel	CirclePixel
pix_area	✔	✖	✖	✖
pix_in	✔	✖	✖	✖
pix_to_mask:all	✖	✖	✖	✖
pix_to_mask:any	✖	✖	✖	✖
pix_to_mask:center	✖	✖	✖	✖
pix_to_mask:exact	✖	✖	✖	✖
pix_to_shapely	✖	✖	✖	✖
pix_to_sky:affine	✖	✖	✖	✖
pix_to_sky:full	✖	✖	✖	✖
pix_to_sky:local	✖	✖	✖	✖

Test	CircleSky	RectangleSky	EllipseSky	PolygonSky
sky_area	✖	✖	✖	✖
sky_in	✔	✖	✖	✖
sky_to_pix:affine	✖	✖	✖	✖
sky_to_pix:full	✖	✖	✖	✖
sky_to_pix:local	✖	✖	✖	✖

[cdeil]

@astrofrog - Can you please rebase this to trigger tests to run again.

Also -- what's the plan here?
Can we merge this now / soon or is this some long-term WIP PR?

[cdeil]

@astrofrog Do you have time to finish up this PR in the next two days for the 0.1 release? Or should we punt this one to the 0.2 milestone?

[astrofrog]

@cdeil - I've just rebased and now made sure that methods that return NotImplementedError are xfailed for now.

Note that there are some genuine failures, either due to my lack of understanding of the API for certain tests, or real issues with different methods. I don't have time to look at this in the short term, but if someone else could take a look and do a PR to my branch to fix the real failures, then I can merge that in and we can make sure this makes it to 0.1. Some of the failures I saw are definitely of the kind that we should investigate before any release.

[cdeil]

@astrofrog - I agree it would be very good to get this in for 0.1. I'll have a look at the fails now and try to fix them.

[cdeil]

@astrofrog - Locally this makes all tests pass for me:
astrofrog#1

I guess the main question is if / how PixCoord should support array values and indexing, but that should be a separate issue / PR?

One more question: what about the "Table of doom" at the top? Does it make sense to copy it to a separate issue?
Probably we want some kind of such overview table concerning the implementation status for certain regions, right?
But maintaining it as a Markdown table is extra work.
Maybe move it to the wiki where everyone can update it and make a reminder issue?

[astrofrog]

I guess the main question is if / how PixCoord should support array values and indexing, but that should be a separate issue / PR?

IMHO it should do what SkyCoord does behavior wise? (that is, it should be able to represent scalar and vector coordinates). I can't remember if SkyCoord actually stores everything as a numpy array though, I think it does.

[astrofrog]

I agree moving the table to the wiki makes sense - can try and do that later today

[cdeil]

travis-ci tests passed.
👍 to merge.

[cdeil]

The Appveyor build fails because shapely is not available in this conda channel:
https://ci.appveyor.com/project/Astropy/regions/build/1.0.74/job/wfb95tev3c94p9jh#L81

@astrofrog - Do you know how to fix this? If no, please remove shapely here and let's move the Windows shapely install issue to a separate future issue.

[cdeil]

@astrofrog - Could you please merge this PR soon?

I'd like to do some more PRs on regions and then do the 0.1 release very soon.
We need it for Gammapy, where we have removed gammapy.regions and are now using the circle regions from this package, i.e. it's a dependency I'd like to have available before making the Gammapy 0.5 release.

[astrofrog]

@cdeil - I'm overloaded today and tomorrow - since you have merge privileges, can you merge this and make the fixes you need to make in a separate PR?

[astrofrog]

@cdeil - actually let me push a fix for shapely, can do it now

[cdeil]

Merging this now. Thanks, @astrofrog !

I'll continue with regions tomorrow.