Just remember to generate separate CA's and certs for the different OVPN instances and clients respectively.
Depending on the number of clients per instance, it might be quite tedious to do client overrides but you should at least do it for the servers.
If you use the internal user manager and generate the certs properly (the CN of the cert should match the username), you should be able to check the logs to determine who has logon to the VPN.
While it is possible to setup pfSense without a LAN interface from 2.X onwards, I would recommend still having a LAN interface for management. Otherwise, pfSense would allow management access on WAN - not a good thing to have this exposed to the interwebs.
As for the multiple instances, once you have tag each instance with an interface name, you can simply regard them as being additional interfaces on pfSense. That is, they behave just like additional local networks on pfSense except that they don't exist physically.
Since these VPN connections are meant strictly for users to connect to your servers, you should make sure not to redirect the gateway (route all traffic through the VPN). In which case, you do not need to worry about NAT rules since all traffic is 'local' to pfSense.