Consider updating the doublet module to run on filtered files

We previously implemented the doublet detection module to run on processed files. However, this might not have been ideal since the method recommends to run on objects that have not been filtered beyond removing empty droplets. We therefore might consider updating that module to run on filtered instead.

That said we're also adding doublet detection into `scpca-nf` which is where we will encourage people to find this information in this future, so this workflow update may not be crucial.

Start by reviewing the doublet detection module's current input and the distinction between processed files and files filtered only for empty droplets. Then compare the value of this workflow change with the planned doublet detection in scpca-nf. Done means a clear decision on whether the update is still needed and, if pursued, verification that the module uses the intended files.

Written by the indexing model from the issue text.