expose disk_space_gb for long-running augur tasks; raise default 375 -> 750

[tomkinsc]

This exposes disk_space_gb as a configurable input for three long-running augur tasks (augur_mafft_align, draft_augur_tree, and refine_augur_tree). This also increases the default from 375 -> 750 for these tasks to increase I/O throughput, as instances as inherently multi-tenant systems (larger disk=fewer parties wanting I/O).

[dpark01]

For GCE local disk, I don’t know if that statement is true about I/O.. each local disk (at 375GB a piece) has a flat I/O rate before they get bound together in a logical volume.

That said, I highly doubt this actually needs to be local instance disk (we’re not processing raw reads or running a kraken db). Normal mounted hdd volumes are probably fine?

[tomkinsc]

This is at the request of @cmloreth based on lessons learned from the Cromwell team; I'll let her explain?