Update initial earliest_exit_epoch calculation & add tests

[jtraglia]

This came up when discussing an edge case: what is the correct earliest_exit_epoch value given a state where the greatest validator exit epoch is less than the current epoch. For example, given a state with one validator and its exit epoch is 3, current epoch is 100, should earliest_exit_epoch be 4? That doesn't seem right.

I believe it should be at least compute_activation_exit_epoch(get_current_epoch(pre)) + 1. And updated to validator.exit_epoch + 1 if there is a validator with a greater exit epoch.

[ralexstokes]

given the implementation of the exit churn computation post-electra, this PR will have no discernible change

I can see this the intent of upgrade_to_electra so fine with me if we want to be explicit about it