Optimize the performance of series migration

[matthias-ronge]

The migration function for series and multi-volume works is very slow. It processes an average of 1000 processes per hour. Just checking whether a process folder contains the anchor file (and if not, skipping it) takes a whole second. This is because services are used through the service loader, it is all very complex. There was also the impression that the existence of files was repeatedly checked many times.

Goal: Review functionality and improve performance. Do more things on the spot, remove everything that is unnecessary to increase processing speed.

[andre-hohmann]

@matthias-ronge: Does the problem still exists?

The SLUB Dresden faced the same problem, but it works now.

[matthias-ronge]

By now, I had no chance to test it again.

Performance issues typically involve various factors, so judging this is always subjective. If you can confirm that a recent hierarchy migration of the SLUB data is going ahead in a timely manner, we can also close this and, if it turns out badly again, open it again.