[mlir][VectorOps] Implement canonicalization for TransposeOp.

[grosul1]

Summary

[mlir][VectorOps] Implement canonicalization for TransposeOp.

Two back-to-back transpose operations are combined into a single transpose, which uses a combination of their permutation vectors.

[bondhugula]

mlir/lib/Dialect/Vector/VectorOps.cpp

1579

Nit: no need of the ref (all derived ops are just one pointer size).

1595–1599

Use dyn_cast_or_null ?

auto parentTransposeOp = dyn_cast_or_null(transposeOp.vector().getDefiningOp());
if (!parentTransposeOp)
  return failure();

[grosul1]

Address Uday's comments

[grosul1]

mlir/lib/Dialect/Vector/VectorOps.cpp

1595–1599

Smaller is better, thanks!

[grosul1]

Existing code uses both styles, but the build fails for the new code. Change
get_permutation -> getPermutation and compose_permutations -> composePermutations.

[rriddle]

mlir/lib/Dialect/Vector/VectorOps.cpp

1571

Please use /// for top-level comments.