eSNI isn't finished and time resolutely insists on moving in one direction, so a document which says "Use this thing that will exist in the future" isn't useful today.
Hasn't it already been implemented in Firefox?
there is an implementation in Firefox (although I don't know if it matches the current state of the spec draft) that unsurprisingly is gated behind an about:config flag.
isn't finished => the standard of it isn't finished so implementations might not to change quite a bit over time and might not be compatible with each other
Currently up to draft-ietf-dnsop-svcb-httpssvc and draft-ietf-tls-esni They're designed more slowly than HTTP3.