Vtable symbol mismatch causes unexpected success

[vecchiot-aws]

This issue should not be resolved until all tests in this PR pass when fixme is removed from the name.

When running rmc <filename> on any of the tests added in this PR, we expect a FAILURE, but get SUCCESS.

This seems to be caused by two things:

1. When CBMC encounters a function which does not appear in the vtable, it (we believe) translates to assume(0).
2. For dynamic traits, different naming conventions are used for insertion and lookup on the vtable.

When specifically running fixme_boxtrait_fail.rs, we receive the following log:

WARN rustc_codegen_llvm::gotoc::rvalue Unable to find vtable symbol for <Concrete as Trait>::increment. Using NULL instead

and when we view the symbol table using rmc --keep-temps fixme_boxtrait_fail.rs and opening fixme_boxtrait_fail.json, we find the following:

...
    "_ZN76_$LT$fixme_boxtrait_fail..Concrete$u20$as$u20$fixme_boxtrait_fail..Trait$GT$9increment17ha9f75436a1c6f32dE": {
      "baseName": "_ZN76_$LT$fixme_boxtrait_fail..Concrete$u20$as$u20$fixme_boxtrait_fail..Trait$GT$9increment17ha9f75436a1c6f32dE",
...
      "name": "_ZN76_$LT$fixme_boxtrait_fail..Concrete$u20$as$u20$fixme_boxtrait_fail..Trait$GT$9increment17ha9f75436a1c6f32dE",
      "prettyName": "fixme_boxtrait_fail::{impl#1}::increment",
...

Note the difference in names used: <Concrete as Trait>::increment vs fixme_boxtrait_fail::{impl#1}::increment.

[avanhatt]

I have an in-progress fix for this that works in all of the fixme_ cases, but breaks a few other tests. Will make a PR once I unbreak those.