Announced masternode blinding - not delivered.
Announced on chain anon (ring sigs) - not delivered.
Those are two off the top of my head, but I'm sure there are others.
Regarding the bolded part, Q1, early Q2 of 2014, the situation with darksend was pretty basic. 10 DRKs inputs in threes, mixing while sending (and having to wait).
The "V2" / second version would be a huge upgrade. Now for V2 there were 2 options of ring sigs and premixing+multiple rounds+multiple denominations (not only 10 DRKs). V2 delivered a radical upgrade through the second way.
I said quite clearly that there isn't anything wrong with changing direction. But it is clear that Evan stating he's going to do X does not mean that what will eventually be delivered actually does X, or even will be delivered at all in every case, as Microchip had suggested with his incomplete list of promised features that included only the ones that happened to be delivered.
The relevance, clearly, to Evolution, is that evolution is quite likely to, "evolve". It can only be properly evaluated once delivered.
*One could argue that there is TOR compatibility.
There is not Tor compatibility with masternodes. They must have a public IP. If you even try to access that IP via Tor (I don't know if there is support for this in the Wallet) you are susceptible to malicious exit nodes. Not recommended (at least not without a very careful review of the protocol to establish what vulnerabilities are introduced). Allowing masternodes to publish an .onion address instead of or in addition to a public IP would fix this, but would perhaps create other issues and I don't see it happening any time soon in any case, as development seems to be focused elsewhere (again stated without any suggestion that this is a bad thing).