Refine existing reproducibility tests

[aidanheerdegen]

Follow up existing MOM5 (or MOM6, they share similar FMS underpinnings) codes for extracting reproducibility information from model output logs.

Establish what is useful, extract/refine as necessary to produce simple standalone tests for reproducibility. These will be incorporated into ACCESS-NRI/model-config-tests#83 but need to tested first.

It may be sufficient to only test MOM5 for reproducibility, as the ice and ocean are very intimately linked. However it may take longer to propagate changes or instabilities from one model into the other, so it would be good to have similar tests for CICE5, but this is a lower initial priority.