Ensure tests fail on stderr and pipe errors

[Syeberman]

Currently, the munit process exits with a success code, even if there is an error creating the buffer for stderr or an error creating the pipe to the forked process. This PR ensures that report.errored is incremented on both those errors, ensuring the process exits with an error code.

Note that the appropriate report counter is currently incremented everywhere else the test result is updated:

munit/munit.c

Lines 1205 to 1226 in fbbdf14

	if (MUNIT_LIKELY(result == MUNIT_OK)) {
	report->successful++;
	#if defined(MUNIT_ENABLE_TIMING)
	report->wall_clock += munit_clock_get_elapsed(&wall_clock_begin, &wall_clock_end);
	report->cpu_clock += munit_clock_get_elapsed(&cpu_clock_begin, &cpu_clock_end);
	#endif
	} else {
	switch ((int) result) {
	case MUNIT_SKIP:
	report->skipped++;
	break;
	case MUNIT_FAIL:
	report->failed++;
	break;
	case MUNIT_ERROR:
	report->errored++;
	break;
	default:
	break;
	}
	break;
	}

munit/munit.c

Lines 1385 to 1386 in fbbdf14

	report.errored++;
	result = MUNIT_ERROR;

munit/munit.c

Lines 1431 to 1432 in fbbdf14

	result = MUNIT_FAIL;
	report.failed++;