Multiple (aliased) "imports" of types when permuting a domain

[rhinomode]

This is a request for an extension to the domain permutation functionality.

When permuting a domain, I would like to be able to narrow a type (with-exclude or with-include) differently depending on where I'm using it in my AST definition (domain). Currently I can narrow a sum-type's variants only at the global/top level. This prevents me from re-using sum-types in another domain in those cases where I want to allow distinct subset of variants in different places (variants) of my own.

Below is a little cooked up example of what I'd like to be able to do:

(define common_ops_ast
  (domain
    (sum core_expr
      (op_a param::symbol)
      (op_b param::symbol)
      (op_c param::symbol))
  ) // end domain
) // end define

(define reuse_ops_ast
  (permute common_ops_ast
    (with_alias core_expr_ops_ab
      (with core_expr    // <------------+-- 'with-alias' allows re-import
        (exclude op_c)))         //      |   of the same domain sum-types with
    (with_alias core_expr_ops_bc //      |   separate (narrowed) permutation
      (with core_expr    // <------------+   of variants
        (exclude op_a)))
    (sum node1
      (op1 expr::core_expr_ops_ab)) // <-- accepts only op_a & op_b from core_expr
    (sum node2
      (op2 expr::core_expr_ops_bc))  / <-- accepts only op_b & op_c from core_expr
  ) // end permute
) // end define

My motivation is specific to PartiQL and its AST; the partiql_ast domain currently has an expr sum-type that includes a ton of useful and handy "operators"/variant which I would like to re-use. I would like to re-use them because I do not want to diverge from the expressions in the expr of partiql_ast. I would like to narrow them as much as possible early on (and relax them later) because I want the PIRG domain I create (by permuting partiql_ast) to present to readers exactly what my backend is capable of (as opposed to me throwing run-time errors when I get something from expr that I haven't implemented or haven't figure out if I want to support).

Thanks,
Chuck