Change the proof of work algorithm

As discussed here: https://mastodon.xyz/web/statuses/99174770035035196

It seems that the current PoW chain (SHA256(Ed25519_sign(SHA256("InnerHash:...Nonce: <nonce>")))) is still ASIC compliant, in that an ASIC could still handle this computation.

@aeris suggests to either use ARGON2 or scrypt currently, because these functions introduce a tradeoff between CPU and memory.

I'm not an expert, so I can just believe this assertion right now.

Anyway, as long as the number of ASIC powered member nodes is low compared to the number of non-ASIC powered member nodes, it won't be a big problem. But it would still be annoying, even more later if Ğ1 growth.