use ClientPool to prevent race conditions when using pylibmc as memcached package

[drcpu-github]

This patch aims to fix concurrency issues when using the gthread worker class in gunicorn in combination with the pylibmc Memcached backend by utilizing a ClientPool rather than a single Client. This issue was also (partially) detailed in https://github.com/pallets-eco/flask-caching/issues/113.

The proposed solution should work for every gunicorn worker configuration, so this patch always uses a ClientPool in an attempt to not further complicate the code with unnecessary branching statements. If you prefer it otherwise, I can still change that using a configuration switch.

Note that this patch also requires some changes to the flask-caching repository, but I first want to iterate on it here before opening another PR in that repository.

Checklist:

• [ ] Add tests that demonstrate the correct behavior of the change. Tests should fail without the change.
• [ ] Add or update relevant docs, in the docs folder and in code.
• [ ] Add an entry in CHANGES.rst summarizing the change and linking to the issue.
• [ ] Add .. versionchanged:: entries in any relevant code docs.
• [x] Run pre-commit hooks and fix any issues.
• [ ] Run pytest and tox, no tests failed.