What you've described is not a "solution" at all - it's just a series of hacks. I'm not a purist and I accept the need for some hacks (e.g I recognize some people will still want to use NAT with IPv6) but the future you're describing, where end users won't have routable addresses, and where it will be difficult/costly to get IP addresses for new services, is totally unacceptable. Transitioning to IPv6 is well within what humans are capable of, and frankly I find comments like yours extremely regrettable because this sort of sentiment is partly why we can't have nice things.
You're also underestimating the difficulty of deploying some of your proposed "solutions." DNS SRV has seen no adoption outside of a few services like SIP and XMPP. Getting web browsers, and every other last client out there that speaks HTTP, to support SRV won't be easy. Just look at the glacial pace other web standards move at.
And factoring in the port number doesn't give you 65536 times as many "addresses." Things like routing and ARP happen at the address level. Under your "solution" you wouldn't be able to migrate/fail over services just by moving their IP addresses between hosts. DNS-based solutions have never worked well for this.
You're also underestimating the difficulty of deploying some of your proposed "solutions." DNS SRV has seen no adoption outside of a few services like SIP and XMPP. Getting web browsers, and every other last client out there that speaks HTTP, to support SRV won't be easy. Just look at the glacial pace other web standards move at.
And factoring in the port number doesn't give you 65536 times as many "addresses." Things like routing and ARP happen at the address level. Under your "solution" you wouldn't be able to migrate/fail over services just by moving their IP addresses between hosts. DNS-based solutions have never worked well for this.