Simplify image options

[zachberry]

Based on a discussion with @iturgeon . Right now images present you with some non-intuitive options. When creating an image you get a few sizes:

• Small (How big is small?)
• Medium (How big is medium?)
• Large (How big is large?)
• Custom (More options?)

The small, medium and large sizes are in actuality just set widths. Small is 18em, Medium is text-width, and Large is maximum width (same width as questions). Therefore, Small is small on a desktop, but on a phone the image can still be the same width, in which case it looks big, so the size label doesn't feel 100% accurate. The sizes can also scale up a picture, so if you upload an image that's 200x200 it may end up scaling up to something like 240x240, resulting in a blurry image. This could be frustrating to authors who upload a clean image and the application gives them an image at nearly the same size, but blurry.

Custom lets you define a width and height, which is tricky. It seems like you need to set both width and height, and if you do that (but give values that don't match the aspect ratio of the image) you'll get a distorted image. Truthfully you can set just one dimension, and the other will be set automatically by the aspect ratio. Typically authors would want to set just the width, allowing the height to be set automatically. However, you can define the height only if you want to, which is uncommon. Another wrinkle - setting dimensions larger than the available content area can distort the image. And another counter-intuitive feature: If you choose custom but don't set the width and height then the image will show at its native size. Imo this is what most authors will want, but I doubt most authors know this option exists.

At any rate, currently images are not "set-it-and-forget-it", there's lots of traps in the various options that likely don't do exactly want you want.

I think this is a step in the right direction: #1669 - This allows you to see the different image sizes as you change them.

That aside, I think we should focus on changing the default behavior. I think when uploading an image it'll default to its native size, which will be the best looking option for the image and likely what authors expect. If we want sizes, perhaps these become percentage-based. So small (for example) is 50% of the content area - Then the image will look small on both a desktop and a phone. If we provide a custom option, then this may be in percentages as well.

Finally, in the future I think we should have a full-width "hero" style image that spans the whole document (which will require #346 )