just running rg0 on node0 and rg1 on node1 is active/active... granted, you normally don't have data traffic in rg0, but you are using node0 for control/routing and node1 for data, so that is active/active. additional redundancy groups can be split between the two nodes (active/active) or you can force all redundancy groups into the same node (active/passive). the 650 cluster doesn't really care, so it really doesn't have a 'switch' to say 'active/active' or 'active/passive'.
based on failures, monitoring, or whatever, you can have an operational cluster switch between active/active and active/passive just as the normal run of things.
if you want active/passive as the standard, just program all the redundancy groups to prefer the same node over the other or use the junos scripts to have rg0 follow rg1. that way, either node will run all rg's, but rg0 will follow rg1 when it can.
if you want active/active as the standard, just program the redundancy groups to run on separate nodes, and turn on preempt (well, don't preempt rg0).
if you really don't care, one way or the other, don't both with preempt and split up the rg's or not -- your call.
basically, active/active or active/passive isn't anything that really applies. operationally, the firewalls are logically viewed as the same box and there is only one control node (whatever that node is), so why care? if you are worried about overloading the fabric link with traffic, or wanting to ensure to get the best throughput, sure -- care, but that is a design issue of reth interfaces, redundancy groups, external hardware, etc. the fact the the srx cluster doesn't really care if running active/active or active/passive is just some nice frosting on the cake and you deal with the real work separately from that.