ci-license-scan: A control char is gibberish enough

Assume that any string containing a control character is gibberish.

Rely on the `unicodedata` module to know which character is a control
character and what is not.

This addresses some bad interactions with control characters ending up
in the diff output, like u+0098 aka Start-Of-String (SOS) preventing the
output to be displayed on terminal emulators like xterm.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

apertis-package-source-builder/overlay/usr/bin/ci-license-scan

@@ -15,6 +15,7 @@ import sh
 import sys
 import yaml
 import textwrap
+import unicodedata
 
 # this is necessary to eliminate references in the generated YAML
 # Perl tools use YAML::Tiny which doesn’t support references.
@@ -226,6 +227,8 @@ def is_gibberish(s: str) -> bool:
     False
     >>> is_gibberish("2000 李健秋")
     False
+    >>> is_gibberish('Æf-w¬6f*äIy,2bÓñ.\x982§VS')
+    True
     >>> is_gibberish('ÿy\x8d\x8aÿt}kÿoiEÿiP')
     True
     >>> is_gibberish('f%^2<')
@@ -235,6 +238,8 @@ def is_gibberish(s: str) -> bool:
     # See https://phabricator.apertis.org/T6677 for details
     if len(s) < 6:
         return True
+    if any(unicodedata.category(c)[0] == 'C' for c in s):
+        return True
     gibberishness = sum(1 if (c >= '\x7b') and (c <= '\xff') else 0 for c in s)
     if len(s) < 16 and s.count(' ') == 0 and gibberishness >= 6:
         return True