[query] Preserve field order in StructExpression.rename

[chrisvittal]

Either set construction from an iterator or set difference is not order preserving. If we iterate through the struct fields in order, then our select inside of rename will preserve order for existing fields.

Brief description and justification of what this PR is doing.

Security Assessment

Delete all except the correct answer:

• This change has no security impact

Impact Description

Query only. Simply causes a single iteration to happen in a different order.

[patrick-schultz]

Oof, good catch! Would you mind adding a simple test to enforce that rename preserves order?

[cjllanwarne]

Looks correct to me, and I agree with @patrick-schultz 's request for a test.

It looks like the rename still shuffles all the renamed fields to the end of the list, even if the original fields are left in their original order. Is that the intended semantics?

[chrisvittal]

We don't necessarily make guarantees as to the order of the fields, but rename is implemented as select which puts 'new' fields at the end of the new struct.