TST: Add exclude list to unicode-check

unicode-check was failing on some unicode characters that are included
in tests for pybind11, which is a submodule of PRIMA. At first I tried
to prevent submodules of PRIMA from being included, but apparently an
explicit call to `git submodule ... --recursive` ignores
fetchRecurseSubmodules (and this explicit call is done by the github
checkout action when it is asked to fetch submodules recursively).

Author: nbelakovski

tools/unicode-check.py

@@ -14,6 +14,12 @@
 extra_symbols = set(['®', 'ő', 'λ', 'π', 'ω', '∫', '≠', '≥', '≤', 'μ'])
 allowed = latin1_letters | box_drawing_chars | extra_symbols
 
+# Some submodules have unicode characters in their tests. Since this is
+# unrelated to scipy we exclude them.
+exclude_directories = [
+    'scipy/_lib/prima/python/pybind11'
+]
+
 
 def unicode_check(showall=False):
     """
@@ -26,6 +32,8 @@ def unicode_check(showall=False):
     for name in chain(iglob('scipy/**/*.py', recursive=True),
                       iglob('scipy/**/*.pyx', recursive=True),
                       iglob('scipy/**/*.px[di]', recursive=True)):
+        if any(excl_dir in name for excl_dir in exclude_directories):
+            continue
         # Read the file as bytes, and check for any bytes greater than 127.
         with open(name, 'rb') as f:
             content = f.read()