We ran into this exact same issue with AWS. In order to get more IPs, we had to add another external interface. It's a pain... There was a lot of tinkering involved, a lot of swearing and a lot of headdesks...
What I had to do was setup policy based routing on the CheckPoint to make sure that incoming and outgoing traffic went in/out of the same interface. I also had to setup incoming/outgoing NAT. Unfortunately, this doesn't work if you are using a Logical Server object to NAT to the ELB CNAME. So we ended up with NAT using the local ELB IP addresses which are subject to change, and when they do, the site goes down...
One of the reasons we needed so many IPs was that ELB's only supported a single certificate. With the new ALBs, they support multiple so if you have a lot of different websites requiring https, you can add multiple to ALBs. So we were able to merge a lot of load balancers and lower the number of IPs we needed.
Also, as Dameon stated, we are in the process of moving part of our stuff to another CheckPoint so that we can get down to 1 external interface and re-implement the Logical Server workaround so we can NAT to CNAME and not have the issue with the ALB IP changing.
Hope this helps and good luck!