The best solution I can see for practically everything you mention is Namecoin/BitDNS for Nostr names (xxxxx.bit). The protocol is under-utilitized but this is a blessing since it reduces the load clients and relays need to run. If clients packaged a light node with their offering and relays ran full nodes (~6.8Gb), Nostr would simultaneously get a global consensus mechanism for: 

i) Human readable/uncensorable names 
ii) Proof-of-Work attached to those names (reducing the potential for spambots to operate without cost)
iii) Ability to register names anonymously 
iv) Deep integrations with Tor for both domains and TLS
v) A key management system so users can have different keys sign different notes (and avoid burning your nsec)   

and 

vi) A far better way to discover and build a social graph than Web of Trust can ever provide on its own

Yes, there's a shitcoin attached to it but the relays could handle everything to do with that so that users wouldnt be encumbered by another token (ie. paying for name operations in sats to relays). So think of it like a shitcoin without shitcoinery.

Namecoin is a serious project with significant miner support (~400ExaHash/s). It has kept current with its development despite a severe lack of interest/funding.

Maybe its time Bitcoin embrace its younger brother?