Multiple improvements
Separation of computations
Computation of scores related to an individual mix or Tx0 is now separated from the computation of statistics for a given pool.
It means that it's now possible to run the score
command without running the load
command. The score
command will search the TXID directly in all snapshot files available in the working directory.
Scalability of the algorithm computing statistics for a given pool
This algorithm has been reworked to cope with the increased activity of the pools during last months.
It now relies on the use of a sketch data structure (Hyperloglog++) allowing to speed up the computation while requiring a small amount of RAM. The tradeoff is that anonsets computed for these statistics have a margin of error which, with current setup, should be around 1% (excluding the margin of error resulting from rounding the anonsets to integer values).