Added a first pass through the GEL section in #123
Things to do next time:
Have better data, ideally. I/O times for Zarr are pretty terrible.
Fix alignment in sample_id col
Clarify a few details about partitions etc during conversion
Stress the point that we're only doing this for one chunk of 106. Would be good to explain the point that we're reading in all of POS and doing binary search for find the right range of variant indexes.
Added a first pass through the GEL section in #123
Things to do next time: