[Feature] Add backend support for AutoCrop for automated headshots

[echo6ix]

Proposal summary

Integrate Autocrop into the stash-box backend to automatically crop existing and newly added photos, into consistent headshots.

What is Autocrop

Autocrop is a Python tool, both from the command line and as a Python API, utilizing OpenCV, designed to automatically detect and crop faces from images, centering the output around the largest detected face.

Benefits of integrating Autocrop with stash-box

• Enhanced Visuals. Headshots improve clarity when performer images are displayed in smaller formats, such as thumbnails and dropdown selects.
• User Curation of Crops Via Voting Or Moderator Selection. Not all automated headshots will be of quality, as this is primarily dependent on the source image. Users, moderators, or admins could select which headshots to use and which to discard either through voting or administrative override.
• Consistency. Automating this process ensures consistent cropping for all images into headshots. Users curated headshots will never be as consistent without the strict use of templates and rigid quality assurance oversight of submissions.
• DMCA compliance. DMCA compliance for performer images is not currently a concern on the stash-box side, however head cropped images are more likely to meet Fair Use criteria in the context of a performer database than unmodified images. Implementing this feature would proactively provide a repository of compliant images, should DMCA considerations arise in the future.