Reject non-ASCII characters in test case descriptions

Don't require that all the tools we use to process test outcomes are
Unicode-clean.
diff --git a/tests/scripts/check-test-cases.py b/tests/scripts/check-test-cases.py
index 8a37a97..87a35e4 100755
--- a/tests/scripts/check-test-cases.py
+++ b/tests/scripts/check-test-cases.py
@@ -64,6 +64,9 @@
         results.error(file_name, line_number,
                       'Forbidden character \'{}\' in description',
                       re.search(br'[\t;]', description).group(0).decode('ascii'))
+    if re.search(br'[^ -~]', description):
+        results.error(file_name, line_number,
+                      'Non-ASCII character in description')
     if len(description) > 66:
         results.warning(file_name, line_number,
                         'Test description too long ({} > 66)',