Add F8 dtype definition in f16_f8_f16 gemm instances (#1092)

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_kn_mn_instance.cpp

@@ -16,6 +16,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
+using F8 = ck::f8_t;
 using F16 = ck::half_t;
 using F32 = float;
 

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_nk_mn_instance.cpp

@@ -16,6 +16,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
+using F8 = ck::f8_t;
 using F16 = ck::half_t;
 using F32 = float;