UIDs are not typed

[Manuthor]

A note for the future (will raise this comment in a dedicated issue if you agree): UIDs are not typed. However, we have different kinds of IDs: User IDs, Key IDs... all are strings. The code-base would be improved if those were of distinct types.

Originally posted by @tbrezot in #968 (comment)

[HatemMn]

I have previously faced this, and want to share my insights for anyone taking this over.
First, is it really a problem... or a decision make since the startup of the KMS and that we want to change ?

In my experience, it was quite boilerplate-heavy to "strongly" type those, and had induced a lot of type conversions for what's basically just like, 16 byte strings.I forgot in which part was the code where I did things that way, but most likely the redis_with_findex part of the codebase

My conclusion was that typing those strings is :

1. put a lot of friction for developers, and waste some non-negligeable time for dev
2. make the code way more human readable and a lot less prone to human error and worse: confusion between two "IDs" that are not the same
3. no computational gain, if not a slight degradation due to boilerplate code (which counts as "free code" unless I'm wrong, if you see the idea)