[MLIR][OpenMP] Fix for intermixed TargetDataOp and TargetOp code

[agozillon]

This is a simple suggested fix (might not be the correct fix) for a problem where TargetOp's nested in TargetDataOp's may have some dependencies on operations inside of the TargetDataOp (a host operation, which can contain both device and host operations) via MapOp's or BoundOps.

The general idea is/would be to treat them similarly to declare target nohost/any functions, but this may not be the correct way to handle it, I only have a cursory understanding of the new flow from reading the PR and tinkering someone with a more intimate knowledge of it may be able to come up with a more sufficient fix.