Fix: Replace surrogates in Unicode code points

[Synrom]

This PR fixes a bug discovered by oss-fuzz. You can find the reproducer here.

The program crashes because surrogates in Unicode escape sequences are just ignored.
That leads to a crash when the string containing surrogates is passed to ascii_lower:

def ascii_lower(string: str) -> str:
    """Lower-case, but only in the ASCII range."""
    return string.encode("utf8").lower().decode("utf8")

UnicodeEncodeError: 'utf-8' codec can't encode character '\udddd' in position 3: surrogates not allowed

This PR fixes this bug, by replacing surrogates with U+FFFD REPLACEMENT CHARACTER (�) in the _replace_unicode function. This is corresponding to the spec.

It also adds a reproducing test case. Also the oss-fuzz reproducer is patched with this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.83%. Comparing base (743c6e5) to head (0109175).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #164      +/-   ##
==========================================
+ Coverage   96.61%   96.83%   +0.22%     
==========================================
  Files           3        3              
  Lines         885      885              
  Branches      136      136              
==========================================
+ Hits          855      857       +2     
+ Misses         14       13       -1     
+ Partials       16       15       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wRAR]

Please run pre-commit.

[Synrom]

Sry, I ran pre-commit now :)