Fix unintended cast of large integer registers to floats

The call to pandas.concat in write_output_wrt_goldenrun() causes large
integer registers values to be cast to floats. Since the register values
do not have to be deduplicated, do not process them with
write_output_wrt_goldenrun().

See also Fraunhofer-AISEC#41 and Fraunhofer-AISEC#42

faultclass.py

@@ -525,9 +525,12 @@ def readout_data(
                 for (flag, keyword, data) in datasets:
                     if not flag:
                         continue
-                    output[keyword] = write_output_wrt_goldenrun(
-                        keyword, data, goldenrun_data
-                    )
+                    if keyword.endswith("registers"):
+                        output[keyword] = data.to_dict("records")
+                    else:
+                        output[keyword] = write_output_wrt_goldenrun(
+                            keyword, data, goldenrun_data
+                        )
 
                 if tbfaulted == 1:
                     output["tbfaulted"] = tbfaultedlist