panopoly_test should not depend on panopoly_pages

The thing is, panopoly_test actually does depend on panopoly_pages for the "body" field base, which it's attaching to it's content type. So, we've got two options:

1. Create our own body field base (maybe "field_panopoly_test_body"?) so we don't depend on it getting created somehow, or
2. Adding features_exclude[dependencies][panopoly_pages] = panopoly_pages to the panopoy_test.info file, so Features won't try to add it back later (and assume that something will have created the 'body' field)

At this point, I'd be willing to go with either approach. #2 is certainly the quickest to implement, and one would expect that every distribution will create a 'body' field somehow...

Also, several of the tests still use the Landing Page content type. Should we switch entirely to Panopoly Test Page?

Hrm, well, all tests (except for those that are meant to test panopoly_pages) should be using the step "When I am viewing a landing page" which creates the "landing page" (a page manager page, not a real content type) on the backend in a way that doesn't depend on panopoly_pages. If there are tests that manually click through the form, then those should definitely get changed to that custom step!

What if we add the exclude, and add an install file to panopoly_test. Once installed we can run node_add_body_field and attach Core's body field to the content type.

That way nothing has to be featurized, unless Behat is checking any customizations done (such as translations.)

This removes Panopoly Pages dependency and uses Core body field on node type insert hook!

Sounds like a good plan! But you still need to add the 'features_exclude' so that the dependency doesn't get added back next time someone does drush fu -y panopoly_tests.

Re-roll with exclude added

Looks good, but I want to try it out on Travis-CI first:

https://travis-ci.org/dsnopek/panopoly/builds/34597625

The last test runs passed, but this took more twiddling in order to get it to take with regard to Features. Here is a new patch! I'll run this one on Travis before moving on as well.

Hrm, it helps when it has more than 0 bytes. :-)

Here's the Travis-CI build:

https://travis-ci.org/dsnopek/panopoly/builds/34634713

Hrm. My patch appears to be failing - I think I went too far. :-) Maybe we could try keeping the field instance in Features, but just doing the features_exclude to block the dependency? That way we don't have to do the hack that @mglaman did with adding the body field manually to the type, it'll just connect to whatever 'body' field happens to be present on the site.

Ok, here's a super conservative attempt.

And here is the Travis-CI build:

https://travis-ci.org/dsnopek/panopoly/builds/34636358

Finally passed on Travis-CI! Commited.