AlternativeImage should be in the fileGroup they originated from

[kba]

We agreed that AlternativeImages for a mets:file representing a PAGE-XML should be written into the same fileGroup as the PAGE-XML document.

From: #471

The idea behind the page_same_group parameter is to restrict deleting images to those that are in the same pc:fileGrp as the PAGE-XML document is, e.g. keeping the image referenced in @imageFilename if the PAGE-XML document is derived from another PAGE-XML document. However, I cannot test this, since I don't have any test data that follows this convention we agreed on in the last Tech Call (that AlternativeImages should be written to the fileGrp of the PAGE-XML that references them).

Yes, this is a hen-or-egg problem. I think we could either start a feature branch in core and at least one preprocessing module to test this realistically, or add a pseudo-processor in the test section here (perhaps derived from the new dummy processor) which produces a dummy AlternativeImage via image_from_segment and subsequent save_image_file.

This issue to track progress on that, without blocking #471.

[bertsky]

We had #530 and PRs in all modules – can this be closed?

Or do we wait until ocrd_im6convert, ocrd_pagetopdf and other bashlib processors do the same fix as OCR-D/ocrd_olena#68, or based on #571?

[kba]

AFAICS only bashlib and bashlib-based processors need further work to properly support the convention. So I think we can close and track progress on bashlib in #571.