Samples intermittently failing to find any assigned fragments

### Context

Trying to process https://github.com/AlexsLemonade/refinebio/issues/2183

### Problem or idea

Some samples fail because they do not meet the threshold of min assigned fragments even though they should. This also happens intermittently (sometimes when we re-process an experiment samples that previously succeeded now fail and vice versa).

### Solution or next step

Since it isn't an issue of data quality, it must be a problem with our pipeline. Is our download being corrupted?

Start by reproducing the processing described in issue 2183 and compare samples that pass or fail the minimum assigned-fragments threshold across runs. Investigate the pipeline's download and fragment-assignment stages for intermittent corruption or nondeterminism. Done means the cause is identified and processing consistently handles samples that meet the threshold.

Written by the indexing model from the issue text.