The casing is supposed to retain enough magnetism to start charging.
I have little electrical knowledge. I've heard excess distance between armature and pole shoe blamed for this needing a boost from the battery to start charging, but I don't know.
In the past, with a lazy dynamo like yours and with an electromechanical regulator, I have fitted a car-type charge warning lamp between dynamo D and a switched battery live. With regulator points closed, that let a little current into the field, or something like that and the dynamo behaved itself.
Maybe doesn't apply to solid state regulators. Anyone know?
Sometimes a dynamo just needs a few miles of riding, to start working.