But typhints in Python are just crutches for developers wishing some semblance of typesafety in an inherently weakly-typed language. They need some kind of static analysis tool to actually make use of this.

On Codewars, there is no such static analysis. The input and return types should be given in the description, aka the specification of the task. Typehints in the solution just clutter up the code and make it frankly less readable.

Only first 20 elements will now be tested to avoid exceptions.

(moreover, we already have a kata specifically about electronic configurations (and doing it right))

Valency should be removed too, as valency is outdated (IUPAC prefers to use oxidation state, for example), and oxidation state is both multi-valued, and so complicated to the point you can force a lot of oxidation states out of nowhere.

IMHO electron configuration should be removed from the kata. Thre are too many exceptions and there isn't a simple rule (except hardcoding). There are also elements without a single undisputed configuration (e.g. nickel).