low/high-S option for P-256 (secp256r1) ECDSA signatures?

[snarfed]

Hi! First off, thank you so much for building and maintaining pycryptodome. It's great!

I'm using it to manage keys and signatures for an AT Protocol implementation. They note that ECDSA signatures ~~aren't always deterministic~~ are malleable, and they try to avoid that by requiring signatures with both secp256r1 and secp256k1 curves to use the "low-S" variant, which is evidently formalized in BIP-62.

Are you all aware of these variants? Do you know which variant pycryptodome currently generates, if either? Any chance we could get an option to DSS.new or nearby to specify the variant?

Thanks in advance!

[snarfed]

Relevant discussion from the ATProto Discord chat:

basically, ecdsa signing relies on the signer picking a secret random value "k", which has to be unique for each message that's signed (otherwise an attacker can recover the private key!). and if your k is truly random then your signatures will be non-deterministic (i.e. different each time, even if you're signing the same data with the same key) but the deterministic algorithms ensure that it's a function of the message being signed, and the privkey, rather than actually random and the signatures themselves are a pair of integers, (r, s) and it just so happens that if (r, s) is a valid signature, (r, -s) is also a valid signature and low-s is just a way of making sure that only one of those is actually valid, by arbitrarily saying that the lower of the two s values is canonically correct