Risk introducing nonce reuse about kms

[bgrieder]

Although object updates are unlikely, they may happen, and we should improve this.
Let us use a random Nonce that is stored with an object. As you did, I would also suggest having the uid as part of the DEM authentication, since we want to prevent an attack where values are switched between 2 Uids

[bgrieder]

PR #65 created against main for quick release; single commit to be cherry-picked on develop.

Note: this is a breaking change on the ciphertexts, but since we just introduced FindexRedis and it is meant to be run in (future) TEEs, I do not think it is an issue. It it is, it will be easy to introduce a fallback on the previous decryption mechanism if the old one fails