The workflow does the following:
- Get all persons for which pages are generated (i.e. associated with the main project)
- Get all depictions of type 'Portrait'
- keep depictions that are 'about' any of the pids of the relevant persons
- keep depictions that have distributions
- keep distributions that have characterized_by with predicate==dcat:downloadUrl
- extract the url (i.e. object)
- register the file in the www-from-model dataset using git annex addurl
The depiction is added as 'portrait.<ext>' alongside the '_index.md' page of the
person page being generated. This is done so that 'qri render-record' and the
associated jinja template can take care, separately, of rendering a portrait
if a 'portrait.<ext>' image file is available.
