If your goal is to force all traffic to one IP address for this scenario, I'm not sure why you'd need GTM unless it's to offload the weight of massive request volume. This is possible, but challenging in that you would need to either have duplicate front ends that can handle traffic arriving in both locations and using a shared backend, or perhaps a little simpler but not without its own challenges, extending an L2/L3 segment at the front door of both datacenters and use LTM in an active/standby scenario by datacenter (by system or just by a traffic group basis)
I'd suggest a professional services engagement or plan on a lot of lab time to iron out all the details and gotchas likely to arise.