Should the compiler reject shift by a type-created type?

[vlstill]

In process of fixing #5094, I noticed that there are testdata/p4_16_samples/psa-dpdk-binary-operations.p4 which have shift by a type that is introduced by type:

// dpdk/psa.p4
// ...
typedef bit<32>  PortIdUint_t;
// ...
type PortIdUint_t         PortId_t;
// ...

// psa-dpdk-binary-operations.p4
// ...
    action forward() {
        b.meta = 32w0x1 << c.ingress_port;
    }

By the spec, this should not be accepted (8.23. Operations on types introduced by type)

Values with a type introduced by the type keyword provide only a few operations:

• assignment to left-values of the same type
• comparisons for equality and inequality if the original type supported such comparisons
• casts to and from the original type

The rationale is probably quite simple -- these type (introduced by type) should be distinguishable from their underlying types. In part to actually protect against uses like these, where the type can be accidentally used in an arithmetic operation, possibly yielding unexpected result.

I think we should reject this code, but it is a breaking change that might require P4 programmers to work around it.