Add differential processing of new Dumplings files

[petrs]

Processing of Dumplings files is time-consuming operation (several hours)

Investigate an option how to process and merge only newly added records from Dumplings files (coinjoins, postmix spend)

[petrs]

Processing optimized to take only 1-2 hours (based on extend of visualizations) for WW2

An attempt with storage in sqlite attempted (load_coinjoins_from_file_sqlite()) => only tens of MBs required, but also ~20x slower performance