SPIKE: Investigate if trie keys with variable lengths are safe

Currently we’re making sure that each Key variant have the same length after serialisation. But being able to use the same key variant prefix but variable payload is also appealing to be able to use store Key items could store data, bids could be prefixed with public key bytes etc. We never tried using variable length Keys as historically it was discussed many times and it is considered unsafe, but we never had any test case to prove such keys are unsafe (or safe). Investigate this and see if it’s unsafe, and what it will take to make such keys safe to use.