Hi Lars, I don't believe there is any benefit of this design and as you mentioned the most effective is way is probably to do SDA in certain cases.
However, in some cases, where the table structure or schema may not readily be known (think a data exploration app) you would not necessarily know which tables need to be in SDA to start with.
Another example of this (in my case with metric2.com) I was hoping (and still am) to offer a cloud based version of my monitoring tool which would be "connected" to multiple other HANA instances, in this case it would be unreasonable to configure SDA for each schema or use-case. But, it would be reasonable to request that a small xsjs/script be placed server side as a DAL.
(PS. Sorry to hijack the OP's thread, but I think it does add some context to the challenge).