Pixel scale and plate scale equivalencies - #4987

Looks good, but these are very simple indeed. Was there a particular need that just (pix * platescale).to(u.arcsec) didn't quite work? (I don't mind having it, but ask in part since I've found that the only one I ever use is dimensionless_angles()).

@mhvk - The use case that prompted me in the first case was an ipython notebook where I was constantly doing things like ((10*u.arcsec)*platescale).to(u.pixel) and of course getting an error because it actually was supposed to be /platescale. Then I would correct myself but later do ((10*u.pixel)/platescale).to(u.arcmin) and then get the other kind of error. That's when it struck me that it really is an "equivalency" in the sense that there's no reason for me to remember whether its angle/pixel or pixel/angle. So it seemed perfect for this because it does what equivalencies are supposed to do: make it easier to do unit transformations with minimal mental overhead.

The other reason this is useful is that it allows using set_enabled_equivalencies - that's particularly useful here because a notebook will often be all about analyzing data from a particular telescope/instrument, so you can just add that at the top and always have the pixel<->angle equivalency available for immediate use. I avoid set_enabled_equivalencies in any situation where it's not clear what's desired, but here I see no other reason why you would want to convert pixel to/from angle.

OK, that makes a lot of sense. I'd definitely make the same mistakes in interactive use...
Since this also all looks OK, I'll merge.