[tests] test_runner - check unicode #10222

#10159 added unicode check/cross/circle glyphs to the test results. This is usually fine, but if Python' stdout can't print unicode, then test_runner will throw an error and not print the results:

 File "./test/functional/test_runner.py", line 482, in <module>
   main()
 File "./test/functional/test_runner.py", line 244, in main
   run_tests(test_list, config["environment"]["SRCDIR"], config["environment"]["BUILDDIR"], config["environment"]["EXEEXT"], args.jobs, args.coverage, passon_args)
 File "./test/functional/test_runner.py", line 288, in run_tests
   print_results(test_results, max_len_name, (int(time.time() - time0)))
 File "./test/functional/test_runner.py", line 316, in print_results
   print(results)
UnicodeEncodeError: 'ascii' codec can't encode character '\u2713' in position 104: ordinal not in range(128)

This change only prints the glyphs if stdout supports unicode, and replaces them with a letter otherwise:

with unicode:

without:

First reported by @TheBlueMatt

What about small "v" instead of "P"?

utACK a97ed80f97a7572f084f1381837f96d6f0cbb569

ACK. Fixed test_runner for me.

@paveljanik the idea is for the characters to be as distinct as possible so it's easy to scan down the list and quickly see if there's a line with a different character (the glyphs were originally added for monochrome terminals where text coloring doesn't help). Small v is too similar to small x.

@jnewbery Fair enough. But P is almost similar to o... In this regard, I think that "v" is much better (and it is also similar to "✓").

utACK a97ed80

@jnewbery Hmm, it got merged in the middle of our conversation 8) What do you think about P and o similarity? Just have a look at your second picture...

@paveljanik thanks for taking the time to think about and review this. I see your point about o/P being similar, but I really think that they're more distinguishable than x/v. I'm also more interested in the symbol for failure being distinguishable from the symbol for success. The times this is most useful is when there's a list of succeeding test cases with one or a few failing test cases, so the x symbol should be instantly distinguishable from the P symbol.

Yes, it is optimized for the most frequent use case. Agreed. Thank you for explanation.