`AveragedTimeInterval` sometimes fails by saving at every iteration

[simone-silvestri]

I have a case that is initialized from a checkpoint. Adding an AveragedTimeInterval to the ocean (or the coupled simulation) makes the output writer write at every iteration going out of memory quite quickly. It seems to be a very niche problem because I cannot find a reproducer. I do not have time now to debug, so I just leave this issue here so that the problem is know and I remember it. I will follow up with a MWE and a (possible) solution