Create empty timeseries array and then fill it snapshot by snapshot, rather than trying to compute the whole timeseries at once within the time for-loop.
Other minor changes:
Added NaNStatistics to Project.toml since we use nanmeans
Renamed files
Removed _diff suffix to \chi because it is redundant with _diffusive (and incorrect for _turbulent case)
Added Nusselt number calculation (array-broadasted ratio) of two buoyancy dissipation timeseries
Bug fixes:
Other minor changes: