load_utils.fov_to_ome is prohibitively slow for large cohorts

[alex-l-kong]

Is your feature request related to a problem? Please describe.

There have been use cases where users would like to create corresponding OME-TIFFs for each FOV in a large cohort. This often arises in the need for data ingestion, where OME-TIFFs are often the choice format.

The current schema requires all of the FOVs to be loaded into memory first, then iteratively written. This is prohibitive both from a time and a space standpoint.

Describe the solution you'd like

Instead of loading in all the FOVs at once, iteratively load in each FOV one at a time. In that way, the users can see the FOVs written immediately and there won't be any memory concerns.