BLD: fix ignored path checking in validate_unwanted_patterns.py

The previous behaviour filtered out too many paths: any subdirectory
whose relative path *contained* any of the ignored paths (which could be
arbitrary strings) would be ignored. E.g., if PATHS_TO_IGNORE contained
"foo", all of "./foo", "./spam/foo", "./spam/foo/eggs", "./barfoobaz",
"./spam/foo.py" would get filtered out.

On the other hand, individual files that *did* appear in the PAHTS_TO_IGNORE
were *not* ignored.

Now the behaviour should be a bit more robust. Ignored file pahts can be
specified as relative paths or absolute paths (since they are all passed
through os.path.abspath); any files below a subdirectory included
in PATHS_TO_IGNORE will be filtered out, and so will any files which are
explicitly mentioned in PATHS_TO_IGNORE.

Author: plammens

scripts/validate_unwanted_patterns.py

@@ -459,20 +459,24 @@ def check_file(file_path: str):
                     )
                 )
 
+    def is_ignored(path: str):
+        path = os.path.abspath(os.path.normpath(path))
+        return any(path.startswith(ignored_path) for ignored_path in PATHS_TO_IGNORE)
+
     if os.path.isfile(source_path):
         check_file(source_path)
     else:
         for subdir, _, files in os.walk(source_path):
-            if any(path in subdir for path in PATHS_TO_IGNORE):
+            if is_ignored(subdir):
                 continue
             for file_name in files:
-                if not any(
+                file_path = os.path.join(subdir, file_name)
+                if is_ignored(file_path) or not any(
                     file_name.endswith(extension)
                     for extension in FILE_EXTENSIONS_TO_CHECK
                 ):
                     continue
 
-                file_path = os.path.join(subdir, file_name)
                 check_file(file_path)
 
     return is_failed