Smaller-Scale Route Reflector Design Question

I am re-evaluating our datacenter interconnect and regarding iBGP I have some design questions on using route reflectors.

I currently am using static routing with tracked objects to automate IP failover between our sites. We have BGP edge routers in each site, with Firewalls that have synchronized configurations. If site A is to become unavailable, the tracked object for the Public IPs routed to the firewalls at Site A will point to Site B's firewall. While a good fit for the time when it was put together initially, as we have expanded it's becoming more cumbersome to add and track many static routes and I would like to have the Firewalls participate in iBGP to advertise their public subnets up to the edge routers.

Due to the rules of iBGP, it seems that I should have a Route Reflector cluster set up, so that routes from Site A are learned at Site B and vice-versa.

Would it be best practice to use a dedicated Router or VM at each site to operate as route reflectors, or would a better approach be to configure our edge routers themselves as route reflectors, saving the need for added complexity?

I've put a diagram together here: http://ift.tt/2BsWbf3 showing a generic view of our topology with the two options. Please let me know if I'm at least on the right track here or if I should even be looking at a different direction to help handle the failover.

Thanks!!