Hi there Arnon.
I forgot to update this thread as well. I had a family emergency and am behind. Sorry everyone! cc @helenaax2r
I do have time to work on making new trees today and tomorrow if you would like to do other things in the meantime. It appears that there are quite a few questions related to ghost-tree at the moment, so I need to make sure things are running smoothly in QIIME 2 which likely involves clarifying things on the tutorial.
ghost-tree was developed before the ASV process, that's why the OTU picking exists. It also has a benefit though. It allows you to regroup your ASVs to lower percent identity threshold. Because ghost-tree partially works using nomenclature, this will allow some of the "unidentified" organisms to cluster with organisms that are "identified" taxonomically in the hopes that you will maintain some of their phylogenetic information rather than discarding them.
Example from the tutorial:
Group your rep seqs at 90% similarity.
This handles the abundant ‘unidentified’ organisms.
qiime ghost-tree extensions-cluster \
--i-extension-sequences extension_seqs.qza \
--p-similarity-threshold 0.90 \
--o-otu-map extensions_otu_map_90.qza
I'm not sure what your question is on point 2.
Hope that helps for now. Let me make some trees 
-Jennifer