QIR generation fails when code includes comparison of tuples or arrays

Nobody has claimed this yet.

RCA does not block programs with comparisons of Result tuples or Result arrays, because Results are considered supported dynamic types. However, comparison across aggregates is not supported in partial eval, so it hits an unexpected failure at QIR generation time.

Here's some example code:
```qsharp
operation Main() : Bool {
use (q0, q1) = (Qubit(), Qubit());
let (r0, r1) = (M(q0), M(q1));

// This version succeeds because it does individual comparisons
// r0 == Zero and r1 == Zero

// This version fails because it compares across arrays.
// [r0, r1] == [Zero, Zero]

// This version also fails because it compares across tuples.
(r0, r1) == (Zero, Zero)
}
```

We need to decide to either support destructing aggregate comparison into individual comparisons in RCA or block comparison of aggregates in RCA so the user gets a design time error.

Start by reproducing the Q# example through RCA and follow the failure at partial evaluation and QIR generation. The work is complete when the project has chosen and implemented either aggregate comparison decomposition into individual comparisons or a design-time error for unsupported aggregate comparisons.

Written by the indexing model from the issue text.