ScalesList methods

[teunbrand]

This PR aims to fix #1310.

In brief, ye olde issue suggested to convert the scales_*() functions to ggproto methods of the ScalesList object, which is what this PR does.

I think it has 2 benefits:

• The ScalesList$scales don't need to be passed around, making such calls a little bit more terse.
• For consistency purposes this is nice, as the upcoming ggproto'isation of the guides follows a similar pattern.

[thomasp85]

blast from the past :-)

[thomasp85]

LGTM - can you confirm that no implementations have been changed (the diff is pretty messed up so I don't fully trust my eye)

[teunbrand]

Other than some style things, the only thing I changed is has_default_transform() check that was in the $transform_df() method and applied it to the $backtransform_df() method as well.

I don't know if it is needed, but it would make the two methods more parallel. (as a reminder, you added this in 101c3a6). Happy to revert if you didn't apply it to de backtransform for good reasons.

The diff is horrible, I'd just open the 1st and 3rd commit side-by-side.

[thomasp85]

that was most likely an oversight by me... I also didn't add any unit tests for it by the looks of it - could I get you to add such a test to make sure we don't end there again

[teunbrand]

I'm not quite sure what prompted the has_default_transform() check. I've guessed it would be custom trans_new() or Scale$transform(), but I can't trigger the code to fail if I leave out the check, so I'm unsure what to test for.

[teunbrand]

I'm merging this. If we want to add a test for the has_default_transform() logic, could you open up a new issue with a little bit more information?