[CK_TILE] FMHA BWD Optimization For GFX950 (#2628)

* simplify fmha_bwd_kernel MakeKargs & dq_dram_window

* simply duplicate

* trload pipeline

* Try two-stage

* add prefetch

* optimize & iglp

include/ck_tile/host/device_prop.hpp

@@ -51,6 +51,12 @@ inline std::string get_device_name()
     default: return name;
     }
 }
+
+inline bool is_load_tr_supported()
+{
+    // Check if load transpose is supported.
+    return get_device_name() == "gfx950";
+}
 } // namespace ck_tile
 
 #endif