A lot of messages are printed very fast when running on all chromosomes or many cells that can't really be read.
Could use spdlog to control logging level or reduce the calls to message, looking for cases where whats true of one sample/chromosome is probably true for all of them.