PDF Image Derivatives could be more efficient

[dannylamb]

When an image derivative is made for a PDF, the entire PDF is used to generate the image. Every page is taken and then flattened into one image, so it takes a while and the resulting image is oversized. With a small chunk of code, you can say "just use the first page", which yields the exact same derivative, just smaller and faster.

What steps does it take to reproduce the issue?
Upload a 50 MB+ PDF and you'll get slow and oversized derivatives

Which version of Islandora are you using?
2+

Any related open or closed issues to this bug report?
I've already integrated this with ISLE to an extent. If this gets merged I can drop a branch i've got over on isle-buildkit.

[rosiel]

It'd be nice to have the page number extracted as an option in the System Action (often ETDs etc have boilerplate cover pages). This'll be a separate issue.

[dannylamb]

Closing this and staying tuned to #2168