I've got the first slice of this feature coded up, just needs a bit of tidying.

I figured that it'd makes sense to break the work down into two parts:

1. Done: synchronously regex check and format urls that have a recognisable image url format
2. TBD: asynchronously check urls and format them as images if the mimetype returned is an image

i'll open up a pull request to get some initial feedback and if you'd prefer not to merge until both parts are completes, that's your prerogative