Group code objects by kernel name in perf report summary (#1234)

Break up the gpu::code_object  print to show the actual kernels...

gpu::code_object::add_kernel: 0.646121ms, 5%
gpu::code_object::mul_kernel: 0.623822ms, 5%
gpu::code_object::add_mul_erf_add_mul_mul_kernel: 0.498902ms, 4%
gpu::code_object::mul_add_kernel: 0.478352ms, 4%

src/targets/gpu/include/migraphx/gpu/code_object_op.hpp

@@ -34,6 +34,10 @@ struct code_object_op
                     f(self.output, "output"));
     }
 
+    value attributes() const { return {{"group", group()}}; }
+
+    std::string group() const { return "gpu::code_object::" + symbol_name; }
+
     std::string name() const { return "gpu::code_object"; }
     shape compute_shape(std::vector<shape> inputs) const;
     argument