Using GDC tiled download as fallback when file is unreadable

Problem

Edge case found with file: https://portal.gdc.cancer.gov/files/7d711bf1-5a44-4bea-ae34-1dec9b3bee4f

For some reason, neither openslide nor TiffFile can read it. In TiffFile, we have an empty pages list, and in openslide we have a "no image data" error. Yet the file seems to appear fine on the GDC viewer.

Possible solution

Step 1: check by manually downloading -> maybe it was a download problem and the file was corrupted. There is a md5 hash in the manifest that we can use for that.

Step 2: add fallback mechanism to the sampler: if the file is corrupted, see if we can at least get the mag / MPP properties and determine the target dimensions, so that we can infer the target level from the tile server, then load it from there.