This dramatically improves load times for large, multi-sample datasets destined for HDF5 output. This does not update the default threads or memory usage for the non-HDF5-backed mode of output. For now, we assume that those with large datasets will use HDF5 mode, and smaller datasets can continue using the default mode.
There could be benefits to the default mode for multi-sample settings, but we have not benchmarked this.
This dramatically improves load times for large, multi-sample datasets destined for HDF5 output. This does not update the default
threadsor memory usage for the non-HDF5-backed mode of output. For now, we assume that those with large datasets will use HDF5 mode, and smaller datasets can continue using the default mode.There could be benefits to the default mode for multi-sample settings, but we have not benchmarked this.
Closes #57