Have you looked at pgsql JSONB datatype and functions[0]? A lot of the "dynamic database" ideas can be implemented using the datatype, and then you have "dynamic sql" available on a already robust rdbms.
For one, it requires learning "new ways" to query, and arguably awkward ways. The goal of Dynamic Relational is to only change the features/conventions of existing RDBMS that affect dynamic-ness in order to reduce the learning curve for those already familiar with RDBMS.
The fact that non-relational or quasi-relational dynamic solutions already exist show there's a need for dynamism. And, I agree that SOME changes are necessary to get the target feature(s), such as dynamism, but it's logical to change as little as possible when inserting the features to reduce the learning curve.
Now, one may argue that add-ons to existing RDBMS give dynamic abilities, but as I suggested before, they are awkward and verbose. If dynamism is a primary feature one wants in an RDBMS, then the verbose approach is not desired.
I welcome other suggestions to achieve similar goals, but as of now, the D.R. suggestion seems the best balance. The goals are: dynamism as a primary feature, and a small learning curve for those already familiar with RDBMS. Go!