Hi @LuSanto I am trying to solve the exact same problem (Demultiplexing CDI reads) and I would like to know more about your workaround method.
The only way around was to split the metadata into several files, so each of them only includes a unique F barcode. Then I used cutadapt separately with each metadata file. Finally, I reimported all the .fasq files into a single .qza for downstream analysis.
Did you run the demux-paired command on the .qza file for each metatada file with the unique F barcode?
If so how did you end up with several fasq files? I thought the output of demux-paired is also a .qza
Thank you!