crucible-mir: Properly parse ArrayToPointer casts

[RyanGlScott]

Although crucible-mir had support for translating these sorts of casts (ArrayToPointer casts), it did not parse them properly. This patch ensures that ArrayToPointer casts are parsed as Misc, which allows them to be handled by the code in Mir.Trans.

Fixes #1224.

[RyanGlScott]

After my changes, the maybe_uninit_array_cast test case now fails:

      mem
        maybe_uninit_array_cast: FAIL (1.39s)
          Compiling and running oracle program                           (0.09s)
          Oracle output: [1, 2]                                          (1.30s)
          Crux output: 
          failures:
          
          ---- maybe_uninit_array_cast/984deb76::crux_test[0] counterexamples ----
          [Crux] Found counterexample for verification goal
          [Crux]   test/conc_eval/mem/maybe_uninit_array_cast.rs:7:19: 7:45: error: in maybe_uninit_array_cast/984deb76::crux_test[0]
          [Crux]   Translation error in maybe_uninit_array_cast/984deb76::crux_test[0]: unimplemented cast: Misc
          [Crux]     ty: TyRawPtr (TyArray (TyInt B32) 2) Mut
          [Crux]     as: TyRawPtr (TyInt B32) Mut
          
          [Crux] Overall status: Invalid.
          
            test/Test.hs:123:
            crux doesn't match oracle

Fair enough, since I changed the relevant code which casts from an array reference (or, in this case, an array pointer) to a pointer to look for ArrayToPointer instead of PtrToPtr, the latter of which is lumped under the Misc umbrella. I could add a dedicated PtrToPtr constructor, but the problem is that there are many different evalCast' cases that could now equally apply to both Misc and PtrToPtr. It would be nice to clean up the code to make this sort of thing easier to express (#1223), but I am inclined not to that just now, since this is intended to be a simple bugfix.

As such, I think I will do the simplest possible thing and map ArrayToPointer to Misc. This is a bit ugly, but it avoids the possibility of introducing regressions like the one seen in the maybe_uninit_array_cast test case. I'll defer the larger task of getting rid of Misc to #1223.

[mccleeary-galois]

Seems like a reasonable option for a quick fix.