Refactor WSP population loader to work with Metis files

[coleshaw]

The population matrices in Magma are calculated from the Flow WSP files for each sample, but it currently only works with S3-hosted files. We need to refactor the population loader to use Metis as a file system, now that we've switched over.

[coleshaw]

Also note that the loader relies on the TUBE NAME keyword in the WSP file. I'm not sure what the previous format that the loader expected was. The "new" format should be IPIGBM020_T1_flow_dc after the pipeline is applied, so the loader will have to account for that when calculating the sample name and stain.