Tracking issue for ConstProp MIR opt

[cjgillot]

This issue tracks the regression in ConstProp due to #110719

Regressed optimizations:

• array indexing;
• non-scalar constants;
• assignment of immutable statics;
• creating rvalues with non-scalar types;
• integer-to-pointer transmutes;
• transmutes of scalar enums.

Remaining to do:

• port const-prop-lint pass to this new implementation.

[RalfJung]

Once this is all done we can also hopefully remove the ConstPropNonsense error variant from the interpreter, and just replace those cases by assertions.

Also we can re-try #114552.

[cjgillot]

Closing as we went for another implementation.

[RalfJung]

Can you add some links to what has been done instead?

[cjgillot]

For bookkeeping, this PR replaced legacy const-prop: #116012