Add missing backports from code base v6.2.2

Fix context to thread management
Fix missing code in selftest.c, autotune.c, hashes.c, dispatch.c and backend.c
Use IS_HIP depending code makes it easier for future optimization related to inline assembly calls - instead of using IS_CUDA || IS_HIP
See TODO markers for more optimizations / next steps

OpenCL/shared.cl

@@ -126,6 +126,11 @@ KERNEL_FQ void gpu_memset (GLOBAL_AS uint4 *buf, const u32 value, const u64 gid_
   r.y = value;
   r.z = value;
   r.w = value;
+  #elif defined IS_HIP
+  r.x = value;
+  r.y = value;
+  r.z = value;
+  r.w = value;
   #endif
 
   buf[gid] = r;