Add support for dtype changes and better chunk calc

[hardingnj]

• includes dtype in calculation for chunk size
• allows casting to other dtype
• allows fill na

@alimanfoo I'm a bit worried about the fill_na step. I'm not sure if there is a better way to do this. Line 183 below.

Snakemake + pipeline changes to follow.

[hardingnj]

Have changed the fill_na step. Tests seem to run ok, and am currently applying to all 29 samplesets on the cluster.

Review welcome- would appreciate input from @amakunin with his experience of snakemake

[hardingnj]

I'm hitting an issue in zarr where if I have multiple jobs creating similar groups, e.g 2L/calldata/MQ and 2L/calldata/AD, even when require_group is used.

The error message suggests that 2L/calldata already exists. I guess this is because there is some race condition if they get created almost simultaneously? Not sure if there is a good way to avoid this? It happens very rarely, so I don't think it's breaking. But would be nice to work around if possible.

[amakunin]

Sorry @hardingnj, just saw this one. Do you still need my review on the issue? From the brief look I cannot see any problems with the code