OK, it now starts making more sense to me.
You can have two PPPoE clients, each with its own username, on the same physical interface and the same service provider, but each of the two PPPoE clients will get its own IP address and the Mikrotik will treat them as two distinct WAN interfaces (pppoe-out1 and pppoe-out2 would be their default names if you don’t specify other ones manually). So to distribute traffic evenly among them, you have to use what is called policy routing. Look e.g. at this post how to use connection-marking and routing-marking, and in the example, replace …list of classifying match conditions for handling A… by pcc=2,0 in-interface-list=all-lan and …list of classifying match conditions for handling B… by pcc=2,1 in-interface-list=all-lan if you want to send 50% of the connections of LAN clients via one PPPoE interface and another 50% via the other one.
E.g. for 3:2 distribution, you would use five connection-marking rules rules in total, pcc=5,0, pcc=5,2, pcc=5,4 to assign new-connection-mark=handling-A and pcc=5,1, pcc=5,3 to assign new-connection-mark=handling-B.
To make use of the assigned routing marks, you have to add the following to your routing table:
/ip route
add dst-address=0.0.0.0/0 gateway=pppoe-out1 routing-mark=handling-A
add dst-address=0.0.0.0/0 gateway=pppoe-out2 routing-mark=handling-B
This is to make it work and understand the principle; you can then make some optimisations to that. In fact, it is enough to mark connections for one of the WANs and let the default handling take care of the other WAN, you just need to make sure that the default route through the WAN which will be used by default handling will have a higher priority (lower value of distance) in the default routing table.