This topic is interesting and before I comment on your post I want to say that IMO this building structural system rationality is not widely understood and a lot of empirical assumptions in analysis, design and construction are the norm.
The way I see it is that yes, there must be a continuous attachment both at the top and bottom of the wall to attain the CMU compression strut masonry action, and the tables and formulas by TMS, ACI and ASCE 530 document reflect those assumptions. If there were no continuous attachment, then the wall shear capacity for in plane forces would not apply and I think the loading would resemble more of a column (pier) with uniformly loaded at its jambs (column flanges), in other words, the wall would deform as a column does.
I have been looking for an example where the compatibility betwwen the steel frame drift and the cmu wall in plate deformation is addressed.
Best regards
Rarebug