Optimise overwriting existing HashList / HashArray when loading SSZ

[arnetheduck]

HashList/HashArray cache the hash of the underlying data - for reasons of simplicity, when we load SSZ into an existing instance (via var overloads), we wipe both data and caches then load the data into a fresh instance.

However, for certain lists - validator sets in ethereum in particular - the majority of such loads do not change the data at all or change it only a little (ie add a few entries or modify a small subset of entries).

It would thus make sense to detect when this happens and not clear the cache, instead letting it be - however, this does not make sense for all lists or for known-large-changes so some heuristics need to be in place.

The overhead of comparing data vs simply overwriting + hashing should probably be studied - re-hashing the validator set on ethereum however is known to take up a significant amount of time.