Our current web solutions don’t use a Django backend.
The idea behind the dictionary backend was that it was stored entirely in memory. A file could be imported when the database was opened and saved when the database was closed.
Similar functionality could be provided using an in-memory SQLite database. We use non-persistent n-memory databases in our unit tests.
I was just thinking that if a Django backend already existed, then including the backend as an add-on would be logical… even if nothing currently used it. That way, if someone had a tool already leveraging Django, they could be that much closer when experimenting with interfacing to Gramps data.
Similar to if a Neo4j backend was available (and I don’t know whether that is possible), we might see the Isotammi group have increased reason to experiment with Gramps interfacing to their in-house genealogy application.