Perform tissue segmentation via SectionAligner code

[mruffalo]

Use @rfmurphy and @tCeZ's SectionAligner code for tissue segmentation, to crop large regions of images that don't contain any tissue.

In the simple case in which only a single tissue section is profiled/included, pass 1 as the number of sections to segment. This may or may not correspond to the existence of a GeoJSON file, so "no GeoJSON file" and "single GeoJSON" file should be treated identically here.

[mruffalo]

@tCeZ pushed the SectionAligner code to https://github.com/hubmapconsortium/SectionAligner.

[mruffalo]

Some modifications to the SectionAligner code are required, such as reading the channels and physical pixel size from the OME-TIFF input.

While those updates are in progress, let's mock up this step as copying the input expression file to the output directory. Everything else in the pipeline will still work for development; we'll just be wasting some CPU, I/O, and storage for large empty regions of the image.

[NickEngland]

This is now complete for the num_tissue=1 case.

We still need to add support for GeoJSON for samples with multiple separate tissues.