Move device_command.hpp to tt_metal/impl and split .cpp (#18762)

### Ticket
#18448

### Problem description
Big header with all impl in it.
Header not needed int metal api anymore (prior PRs removed the
dependency)

### What's changed
For now just splitting as is.
Template could go away to make this maintainable.

### Checklist
- [x] [All post
commit](https://github.com/tenstorrent/tt-metal/actions/runs/13713426427)

tests/tt_metal/tt_metal/dispatch/dispatch_util/test_device_command.cpp

@@ -2,10 +2,10 @@
 //
 // SPDX-License-Identifier: Apache-2.0
 
-#include <tt-metalium/device_command.hpp>
 #include "cq_commands.hpp"
 #include "memcpy.hpp"
 #include <gtest/gtest.h>
+#include "tt_metal/impl/dispatch/device_command.hpp"
 #include "tt_metal/impl/dispatch/device_command_calculator.hpp"
 
 namespace tt::tt_metal {

tt_metal/distributed/mesh_trace.cpp

@@ -7,7 +7,7 @@
 #include <mesh_coord.hpp>
 #include <mesh_trace.hpp>
 
-#include <tt-metalium/device_command.hpp>
+#include "tt_metal/impl/dispatch/device_command.hpp"
 #include "tt_metal/distributed/mesh_workload_utils.hpp"
 #include "tt_metal/impl/trace/dispatch.hpp"
 

tt_metal/distributed/mesh_workload.cpp

@@ -8,7 +8,7 @@
 #include <tt_metal.hpp>
 
 #include <tt_metal/impl/program/program_command_sequence.hpp>
-#include <tt-metalium/device_command.hpp>
+#include "tt_metal/impl/dispatch/device_command.hpp"
 #include "tt_metal/distributed/mesh_workload_utils.hpp"
 
 namespace tt::tt_metal::distributed {

tt_metal/distributed/mesh_workload_utils.cpp

@@ -5,7 +5,7 @@
 #include <host_api.hpp>
 #include <command_queue.hpp>
 
-#include <tt-metalium/device_command.hpp>
+#include "tt_metal/impl/dispatch/device_command.hpp"
 #include "tt_metal/impl/program/dispatch.hpp"
 #include "tt_metal/impl/dispatch/dispatch_query_manager.hpp"
 #include "tt_metal/distributed/mesh_workload_utils.hpp"

tt_metal/impl/CMakeLists.txt

@@ -22,6 +22,7 @@ set(IMPL_SRC
     ${CMAKE_CURRENT_SOURCE_DIR}/program/program.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/program/dispatch.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/dispatch/debug_tools.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/dispatch/device_command.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/dispatch/host_runtime_commands.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/dispatch/device_command_calculator.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/dispatch/dispatch_query_manager.cpp

tt_metal/impl/buffers/dispatch.cpp

@@ -2,10 +2,11 @@
 //
 // SPDX-License-Identifier: Apache-2.0
 
-#include <device_command.hpp>
 #include <device.hpp>
 #include "assert.hpp"
 #include "dispatch.hpp"
+#include "tt_metal/impl/dispatch/device_command.hpp"
+
 #include <tt-metalium/command_queue_interface.hpp>
 #include <tt-metalium/dispatch_settings.hpp>