I read the sidebar and I still have an issue with the two archetypes as presented. Names matter, and when the designers use names that have strong connections to real world warriors it matters to me that the mechanics back that up. As is the UA Knight is two concepts in one - mounted warrior and defender/bodyguard - each would be better suited by a separate, dedicated archetype using a generic name.
I suppose the issue for me is this: do classes represent a bundle of related mechanics or do they mean something from and in-world perspective? When designers use names like Knight or Samurai, I expect something more than just a bundle of mechanics that evoke a certain feel. The UA Knight and Samurai do not feel like actual knights and samurai as those concepts are better covered by backgrounds. I have similar issues with the Barbarian (how barbarian should the barbarian be? How do they differ from fighters with the Outlander background?).
It's for this reason that I liked the Fighting Man and Magic User of OD&D: they describe what a class does with minimal baggage.
D&D has, for three editions now, represented knight as a defender with horseman skills. It was in
3e, and it was in
4e. At this point, the idea of knight as defender is as established as warlock = always on magic or barbarian = rage warrior.
Now, if you want to delve into the deep abyss that is "does class represent an in-world concept or not", you're going to get about 100 pages of debate. I'm of the opinion that it does, and that most PCs, while they might not say "I'm a barbarian", it does represent a certain concept or training or possible occupation. To the specifics of if a knight is a background or subclass, I ask "why not both?" I mean, there is ALREADY two knight backgrounds (variant noble and Knight of the Order) as well as the whole concept of the Paladin (esp the Oath of the Crown), so in the short amount of time 5e has been around, we already have a half-dozen ways to emulate a knight. This fighter-knight is just another way to do so.
As for the last point, I can 100% disagree with you on this. Generic names are good for generic classes that are purposefully vague and uninspiring, which is the opposite of D&D. D&D classes are archetypes, and the names should reflect that. Classes like ranger, bard, monk, barbarian, or warlock invoke a certain automatic response that names like "woodman, entertainer, martial artist, wildman, or pactbound" do not. Personally, the "generic name" concept of OD&D died the minute "thief" entered the game, so the point is rather moot at this point.