[mlir][tensor][nfc] Clarify comments for `createPadHighOp`
While reviewing this code, I realised that the rationale behind the
assert was not very clear, so I updated the comments to clarify it.
Relaxing the assert (i.e., allowing `resType.getNumDynamicDims() !=
dynOutDims.size()`) would require generating a mapping between
`dynOutDims` and the dynamic dimensions of the output tensor. At the
moment, this additional complexity is unnecessary.
To minimize PR noise, I am submitting this without a review. However,
please ping me if you believe this or similar changes should be reviewed
before merging.
