Failed persistence load, fallback full reparse does not work as intended #295
Description
The model for msc_file_load is to fall back to a full reparse if there is a failure loading persistence data.
The schema changed from 0.0.8 to 0.0.9 and I forgot to --startclean on a test box I'd previously been using 0.0.8.2 with. This has highlighted that when we fail to load persistence we are not correctly reparsing as intended. The behaviour appears to be a spinning core (ie full utilization) but no progress - after this line:
msc_file_load(C:\Users\zathras\AppData\Roaming\Bitcoin\MP_persist\offers-000000000000000008b15ce3f1f0431e2d37acdce289e89500d41e66c6987522.dat), loaded lines= 0, res= -1
We are returning a negative success (res=-1 during load_most_relevant_state() and then we do not write any additional data to the log during 60 or so minutes of 100% usage of a CPU core (reparsing would result in entries being written to the log).