test_matlabio has stochastic failure

[zm711]

Describe the bug

Added testing for python 3.9 and 3.11 (this is the new io to test) for the io-testing. This led to one failure in doing the roundtrip size test in the test_matlabio. It went away on a rerun. We should watch this to see if this is a stochastic test or was just a CI fluke.

To Reproduce Track the CI

Expected behaviour

We may need to rewrite the test if it has random variations on newer versions of python.

Environment:

• OS: Linux CI
• Python 3.11.x
• Neo version 0.14.0.dev
• NumPy version 1.22.x

Additional context cc @apdavison