The use case is:
1) I have different "flavours" of the same service that is for example a "transparency portal" composed by Postgresql node +tomcat-lifaray node + apache reverse proxy node as a basic topology for a minimal PA, but I could want to change the topology to scale the service for a bigger PA and I cloud have a cluster of tomcat and a couple of Apache in front with a load balancer.

As can be seen these are two different topology that cannot be "just scaled" but need different planned topology.

Logically this is 1 service thus 1 TOSCA file but I have different sizing.
In the XML spec we have that Definition allows for having more than 1 ServiceTemplate and this is logical since the ServiceTemplate contains all the needed elements to allow for different sizing.
Is a proper interpretation of the standard or am I stretching it too far?
I believe that this use case allows the orchestrator to interpret the TOSCA file properly looking at how many ServiceTemplate are there and presenting the end user with a choice of sizing for the service and afterwards the request for the inputs