SWDEV-505971 - change setArgument arg from uint32_t to uint64_t

We are passing this arg as an address, and memcpy complains about
overreading (8 bytes instead of 4).

Change-Id: Ica9207f6c5f6056a4bfc968280c76e779ded13ae

rocclr/device/blit.hpp

@@ -236,7 +236,7 @@ class BlitManager : public amd::HeapObject {
   //! Stream batch memory operation
   virtual bool batchMemOps(const void* paramArray,
                            size_t paramSize,
-                           uint32_t count
+                           uint64_t count
   ) const = 0;
 
   //! Enables synchronization on blit operations

rocclr/device/pal/palblit.hpp

@@ -478,7 +478,7 @@ class KernelBlitManager : public DmaBlitManager {
   //! Batch memory ops- Submits batch of streamWaits and streamWrite operations.
   virtual bool batchMemOps(const void* paramArray,
                            size_t paramSize,
-                           uint32_t count) const {
+                           uint64_t count) const {
     assert(!"Unimplemented");
     return false;
   }

rocclr/device/rocm/rocblit.cpp

@@ -2513,7 +2513,7 @@ bool KernelBlitManager::streamOpsWait(device::Memory& memory, uint64_t value, si
 
 // ================================================================================================
 bool KernelBlitManager::batchMemOps(const void* paramArray, size_t paramSize,
-                                    uint32_t count) const {
+                                    uint64_t count) const {
   amd::ScopedLock k(lockXferOps_);
   bool result = false;
   uint blitType = BatchMemOp;

rocclr/device/rocm/rocblit.hpp

@@ -505,7 +505,7 @@ class KernelBlitManager : public DmaBlitManager {
   ) const;
 
   //! Batch memory ops- Submits batch of streamWaits and streamWrite operations.
-  virtual bool batchMemOps(const void* paramArray, size_t paramSize, uint32_t count) const;
+  virtual bool batchMemOps(const void* paramArray, size_t paramSize, uint64_t count) const;
 
   virtual amd::Monitor* lockXfer() const { return &lockXferOps_; }