chore: Ocsp timeout adjustment

[dougch]

Description of changes:

While investigating a spike in integration test runtimes, I came across a single OCSP test that took 30 minutes, which appears intentional. The average runtime of the whole test suite is 65 seconds.

Dial back the wait time to 90 seconds and add a flaky mark with 3 retires, so if there is a hang, it gets cleaned up faster and retried. If there is more context or recommendation about a better timeout, let's bump it.

285: ============================= slowest 10 durations =============================
285: 2000.05s call     test_ocsp.py::test_s2n_server_ocsp_response[OCSP_RSA-TLS1.2-P-521-S2N-OpenSSL-ECDHE-RSA-AES128-SHA256]
285: 1.08s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-256-S2N-OpenSSL-DHE-RSA-AES256-GCM-SHA384]
285: 1.06s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-256-S2N-OpenSSL-DHE-RSA-AES128-SHA]
285: 1.06s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.1-P-256-S2N-OpenSSL-DHE-RSA-AES256-SHA]
285: 1.06s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-521-S2N-OpenSSL-DHE-RSA-AES128-GCM-SHA256]
285: 1.05s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-256-S2N-OpenSSL-DHE-RSA-AES256-SHA256]
285: 1.05s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-256-S2N-OpenSSL-DHE-RSA-AES128-SHA256]
285: 1.04s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-521-S2N-OpenSSL-DHE-RSA-AES256-SHA]
285: 1.03s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-384-S2N-OpenSSL-DHE-RSA-AES128-SHA]
285: 1.03s call     test_ocsp.py::test_s2n_client_ocsp_response[OCSP_RSA-TLS1.2-P-521-S2N-OpenSSL-DHE-RSA-AES128-SHA]
285: =========================== short test summary info ============================
...
285: ================== 576 passed, 1 rerun in 2012.69s (0:33:32) ===================
1/1 Test #285: integrationv2_ocsp ...............   Passed  2013.06 sec

Testing:

How is this change tested (unit tests, fuzz tests, etc.)? Are there any testing steps to be verified by the reviewer?

Is this a refactor change? If so, how have you proved that the intended behavior hasn't changed?

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[lrstewart]

Hmmm the original PR had 2000, but no explanation for it and none of the reviewers questioned it. @goatgoose any chance you remember the reasoning from almost 3 years ago? :P

[goatgoose]

I do not remember sadly. I'm guessing it was a mistake though. 2,000 seems way too high to be intentional. Maybe leftover from debugging or something.