[DEBUG on CI] Use alternative instructions for converting f16 to i8

Author: xuchen-intel

src/plugins/intel_cpu/src/emitters/plugin/aarch64/jit_conversion_emitters.cpp

@@ -216,7 +216,10 @@ static void jit_convert_process(dnnl::impl::cpu::aarch64::jit_generator* h,
                     cvt_dbyte_to_byte<isa>(h, out_idxs, out_idxs, output_type.is_signed(), is_saturated);
                     break;
                 case ov::element::f16:
-                    cvt_f16_to_dbyte<isa>(h, in_idxs, out_idxs);
+                    // cvt_f16_to_dbyte<isa>(h, in_idxs, out_idxs);
+                    cvt_f16_to_f32<isa>(h, in_idxs, out_idxs);
+                    cvt_f32_to_i32<isa>(h, out_idxs, out_idxs);
+                    cvt_i32_to_dbyte<isa>(h, out_idxs, out_idxs, output_type.is_signed(), is_saturated);
                     cvt_dbyte_to_byte<isa>(h, out_idxs, out_idxs, output_type.is_signed(), is_saturated);
                     break;
                 case ov::element::i8: