[llvm] [OpenMPIRBuilder] Emit a dispatch loop for ordered worksharing loops on the device (PR #214263)

[Spencer Bryngelson via llvm-commits]

sbryngelson wrote:

Ping. @jdoerfert @jhuber6 could one of you take a look?

This is a wrong-results fix: `ordered` inside a target region orders nothing in flang, so a conforming program silently gets the wrong answer. The C equivalent is correct on the same GPU. CI is green.


https://github.com/llvm/llvm-project/pull/214263