Fix for #300 (#331)

On some versions of numpy/pandas, applying a str map to an empty series doesn't change the type.
As a result, when adding a float to a string, sometimes numpy dies with a uadd error.
We fix this by explicitly casting to a str first

Signed-off-by: Nicholas Curtis <nicurtis@amd.com>
Co-authored-by: Nicholas Curtis <nicurtis@amd.com>

[ROCm/rocprofiler-compute commit: d72ef52208]

projects/rocprofiler-compute/src/utils/tty.py

@@ -149,6 +149,7 @@ def show_all(args, runs, archConfigs, output):
                                             .astype(float)
                                             .round(args.decimal)
                                             .map(str)
+                                            .astype(str)
                                             + " ("
                                             + t_df_pretty.map(str)
                                             + "%)"