What is the ideal scenario for using this feature?

I have an interesting nosqlbench use case that implies Linearized operations as stated on devdocs/linearized/linearized.md

Is your feature request related to a problem? Please describe.
In the use case, we are using a clustering column which usually contains 20 values per partition key in a table . Algorithm is :
Partition-key based read on table A
Foreach (ck), Do a select in another table.

I can mitigate the problem for now, since I control data through a rampup phase, I can
reuse bindings for main phase.
I'm just thinking of how I could have done it without "knowing" table data.
I understood that "linearized" is design sketch only for now.
in inject syntax, it is stated :
The variable is assumed to be a single-valued type.
It's a logical limitation considering presented capture syntax.

I like the simplicity of the proposed syntax and don't think it should be changed.

I think this problem could be solved using, somehow, specific configs on statement or block that would both specify :

• declarative dependency between statement 2 and statement 1
• handling of resultset: first, last, foreach, etc...