I don't have strong documentation, but the general process we use is, in your scenario, would be as follow:
1. Configure the 3600 with the required network settings and as a master-backup to your M3, using VRRP and master-redundancy, but set the VRRP priority lower so that the 3600 doesn't take over for the M3. This will allow all the configuration, WMS DB, and mesh recovery profiles to be pulled over to the 3600.
2. Now you can pull the M3 controller and the 3600 will take over as the master. You can then re-IP the 3600 to the old M3 IPs (assuming you have removed the M3), and change the VRRP priority back to where the M3 was.
The good thing about this is, so long as you're not terminating APs on the master, it's not really a hard outage. If you ARE terminating APs on the master, you will have to take an outage.