1. This sounds like a weird one. It would be helpful to have some more info about your network. Would you share your PVE host’s /etc/network/interfaces file and the config file for the VMs (from /etc/pve/qemu-server)?
2. Excellent
3. I think ZFS would likely help since it can make use of block level snapshots. I think the way to move things over would be to create a ZFS datastore in Proxmox and then just migrate each VM’s disk.
4. Personally I think this is a bit simpler in an LXC container and there are a bunch of tutorials to help. These two are similar to my own setup:

• https://blog.bekh.fr/jellyfin-lxc-with-hardware-transcoding-igpu-passtrough/
• https://www.wundertech.net/installing-jellyfin-on-proxmox/

Hope some of that helps