update ActiveSupport::TestCase#assert_not to be ruby styleguide compliant

[SampsonCrowley]

Summary

• use refute value, not assert !value
• add additional testing for "truthy" values

Other Information

The general ruby style guide says to prefer refute over assert !

The current implementation goes directly against that under the premise that "minitest might change refute" (see here)

1. that is not a good reason to not use refute, in my opinion.
   • Minitest is not going to randomly change one of the two core methods of the framework on a whim.
2. instead of going against the style guide on the extremely slim chance that Minitest completely rewrites their core methods I believe it's better to follow the style guide, and instead add additional testing to check that the values we want are properly accepted or rejected
   • this has the benefit of following the core ruby style guide
   • it also verifies that if a change is made in Minitest, our testing will notify us of that change

[georgeclaghorn]

Thanks for the PR, but (a) this is a cosmetic change, which we don’t accept and (b) we don’t follow the (unofficial) Ruby Style Guide.

[SampsonCrowley]

(a) Sorry, didn't think about that.

As for (b) No, not officially, but the basic rules of the general Ruby guide are well known to all of the Ruby devs I know, and I think it's fair to say those rules that don't violate the rails guide can still have value for recognizability and implicit logic. Because refute not only already exists, but is the preferred way of the authors of the library, the existence of the assert_not method was a curious ocurance without an immediate explanation. It wasn't clear whether assert_not was an alias for or just equivalent to refute or if there was a definitive performance reason for using it or why.

@georgeclaghorn if I update the docs and/or testing guide page to include a link to the rails style guide or note that the method is equivalent to refute would that be acceptable?