storage: write data to ADC filestystem tile store

[mbjones]

Store workflow outputs by writing data and metadata into a filesystem mounted directory provided by the orchestration engine, usually as a mounted PVC in our kubernetes deployments. The directory structure is typically going to contain package-level metadata and a format-specific arrangement of files. For example, a WMTS workflow would arrange files in directories following a Tile Matrix Set (TMS) such as a quad tree, whereas a Zarr dataset would arrange files following the chunking dimensions chosen for Zarr.

Regardless, the dataset is written to a filesystem location (rather than a dataset as described in #70 and #71), and it is up to the ADC or other data center about how to publish and distribute these files.