There is very little information on this out there in regards to if DNS Round Robin being supported for a CAS Array when NLBS isn't available (in such cases where you're using 2 total servers running DAG groups at the same time) and a hardware load balancer isn't available.
The short of it is that this works spectacularly. What few people understand is how DNS Round Robin truly works. The DNS Server itself dishes out both IP addresses associated with the dual record to the client at all times. The difference is in that the DNS Server round robin's which IP address is first when it responds to the client. The client has both addresses in it's local DNS cache at this time.
When this happens, the client tries to connect to the first address and if it does not respond, it fails over to using the second IP address in its cache. Super. This generally occurs within 30 seconds. Additionaly, when the client is connected to a CAS server and that connection drops for any reason, the client then tries to connect to the alternate IP address at that time.
The outage time experienced for Outlook in this scenario is somewhere around 15 to 30 seconds in all my testing.
Now, does this help with OWA? No, you will have to re-authenticate as that session is lost. Is it acceptable on a small budget with little impact in the event of a failure? Yes, by all means.