Report output fails when count matrix has float counts instead of integers

[MNM-TB]

Thank you for a fantastic tool!

I know that mine is somewhat of an edge case, but our research requires careful counting of multimappers i.e., reads from the 10X Chromium pipeline that perfectly maps to multiple loci or overlapping genes. CellRanger does not support those and thus we are using the STARsolo component of the STAR aligner which works perfectly.

However, the Expectation-Maximization (EM) algorithm results in read scouts being stored ads fractions in the read matrix and not integers. While CellBender manages to train on and filter these matices well, the reports fail after the first UMI curve plot with this error:

`

AssertionError Traceback (most recent call last)
Cell In[3], line 1
----> 1 generate_summary_plots(input_file=input_file,
2 output_file=output_file,
3 truth_file=truth_file)

File /media/data/TomasB/pyProjects/snRNA-rat-BC/CellBender/cellbender/remove_background/report.py:154, in generate_summary_plots(input_file, output_file, truth_file, dev_mode)
152 # prove that remove-background is only subtracting counts, never adding
153 if out_key == 'cellbender':
--> 154 assert (adata.layers[input_layer_key] < adata.layers[out_key]).sum() == 0,
155 "There is an entry in the output greater than the input"
156 else:
157 if (adata.layers[input_layer_key] < adata.layers[out_key]).sum() == 0:

AssertionError: There is an entry in the output greater than the input

`

It would be very nice to have the reports if possible.