Update device.h for hip_bfloat16 inclusion guard (#2107)

* Update device.h for hip_bfloat16 inclusion guard

Prevents other files in rocm include the old hip/hip_bfloat16.h, which is guarded by _HIP_INCLUDE_HIP_AMD_DETAIL_HIP_BFLOAT16_H_ and _HIP_BFLOAT16_H_

* Update device.h to handle old hip_bfloat16.h

Added a workaround for old hip_bfloat16.h header usage.

[ROCm/rccl commit: 8e4dbfdf37]

projects/rccl/src/include/device.h

@@ -12,9 +12,16 @@
 #include "nccl.h"
 #include "rccl_float8.h"
 #if ROCM_VERSION >= 60000
-   // hip_bf16.h should be used from ROCm 6.0
-  #include <hip/hip_bf16.h>
-  typedef __hip_bfloat16 hip_bfloat16;
+  // This is a workaround for the fact that the old hip_bfloat16.h header file may still be used by some rocm files.
+  // The _HIP_INCLUDE_HIP_AMD_DETAIL_HIP_BFLOAT16_H_ and _HIP_BFLOAT16_H_ macros are defined in the old hip_bfloat16.h header
+  #if !defined(_HIP_INCLUDE_HIP_AMD_DETAIL_HIP_BFLOAT16_H_) && !defined(_HIP_BFLOAT16_H_)
+    #define _HIP_INCLUDE_HIP_AMD_DETAIL_HIP_BFLOAT16_H_
+    #define _HIP_BFLOAT16_H_
+    #include <hip/hip_bf16.h>
+    typedef __hip_bfloat16 hip_bfloat16;
+  #else
+    #error "RCCL is not using the correct hip_bf16.h file. Please make sure that the correct header is included!"
+  #endif
 #else
   #include <hip/hip_bfloat16.h>
 #endif