[llvm] [mlir] [Offload] Add oneInterationPerThread param to loop device RTL (PR #151959)

[Michael Kruse via llvm-commits]

================
@@ -767,8 +767,17 @@ template <typename Ty> class StaticLoopChunker {
 
 public:
   /// Worksharing `for`-loop.
+  /// \param[in] Loc Description of source location
+  /// \param[in] LoopBody Function which corresponds to loop body
+  /// \param[in] Arg Pointer to struct which contains loop body args
+  /// \param[in] NumIters Number of loop iterations
+  /// \param[in] NumThreads Number of GPU threads
+  /// \param[in] ThreadChunk Size of thread chunk
+  /// \param[in] OneIterationPerThread Assume that one thread executes
----------------
Meinersbur wrote:

```suggestion
  /// \param[in] OneIterationPerThread If true/nonzero, each thread executes only one loop iteration or one thread chunk. This avoids an outer loop over all loop iterations/chunks.
```
[suggestion]

https://github.com/llvm/llvm-project/pull/151959