The Collection mgmt test_create_collection_max_expiry test can crash intermittently in the integration-test-pipeline. It seems very specific to the environment: cbdyncluster + SDK running on x86_64 centos7 machine.

Creating this ticket to investigate. I tried to repro locally at one point, but b/c I am on an M1 I was not able to get the appropriate backtrace. Maybe things have improved by now and/or maybe we can update the CI pipeline to provide the backtrace (quite positive we do it in our C/C++ pipelines, but maybe there is weirdness w/ Python).