Coverage remained the same at ?% when pulling 1fda7a4 on kckaiwei:transform into b8e92e7 on jazzband:master.

Could use a 2nd eye on this. Adding an option changed the key used for caching (since the key is partially made by options supplied), so I updated the paths for the tests.

Can't the change of keys be avoided somehow? A change would be quite bad for big sites with lots of images.

That's a good point. I guess it could go in with the EXTRA_OPTIONS, and thus avoiding the need to recache every image. But the current EXTRA_OPTIONS all felt to be reflected as global variables set in the settings file. Although the docs do have a blurb on setting orientation on a image by image basis so there seems to be a precedent on it.

I'll make the changes accordingly.

Conflicts need fixing

Ensuring resolved conflicts still work, set to WIP

@kckaiwei are we ready to merge this?

No, I actually forgot about this tbh, I don't work at the same place that used it last time, so lemme see if I can get a test running for this, iirc. I had a bunch of failing tests on Travis last time due to it that I couldn't figure out.

Thanks. Take a look and let us know. If you think there is no way to get this up and running we better close this PR