add stage that checks for introduced/fixed crashes

    # remove leading path from testcase names for better readability

    df_merged["testcase"] = [pathlib.Path(tc).name for tc in df_merged["testcase"]]

    # this is for printing the whole testcase names

    pd.options.display.max_colwidth = 200

    regressions_found = False

    # check for newly introduced crashes

    col_curr = "Result-curr"

    col_prev = "Result-prev"

    mask_crash_introduced = (df_merged[col_curr] == "ERROR") & (

        df_merged[col_prev] != "ERROR"

    )

    mask_crash_fixed = (df_merged[col_curr] != "ERROR") & (

        df_merged[col_prev] == "ERROR"

    )

    display_cols = ["testcase", col_curr, col_prev]

    if sum(mask_crash_introduced) > 0:

        regressions_found = True

        print("---------------Testcases that introduced new crashes---------------")

        print(df_merged[mask_crash_introduced][display_cols].reset_index(drop=True))

        print()

    if sum(mask_crash_fixed) > 0:

        print("---------------Testcases that fixed crashes---------------")

        print(df_merged[mask_crash_fixed][display_cols].reset_index(drop=True))

        print()

    # remove columns with ERRORs in any of the csv files before comparing the numerical columns

    mask_no_errors = (df_merged[col_curr] != "ERROR") & (df_merged[col_prev] != "ERROR")

    df_merged = df_merged[mask_no_errors].reset_index(drop=True)

    # check numeric columns and compare diff to thresholds

    for col in args.columns_to_compare:

        col_curr = f"{col}-curr"

        col_prev = f"{col}-prev"