1) True, but it's still a better single-gen increase than Intel has delivered since Sandy Bridge. Zen+ had 2-3% better IPC than Zen. 115/103=1,117, or still nearly 12% improved IPC over Zen+. It's not 15, but it's better than anything Intel has managed.
2) Why would 7nm deliver higher clock speeds? As with all process nodes like this, it's a mobile-first, low power process node. Getting it to 4,6GHz is quite impressive. I guess we'll see about OC headroom, but even 4.6 OOTB is excellent, as it will at least mean most chips can OC to all-core 4.6GHz (unless they already do so).
3) What does branding matter? Sub-$200 6c12t CPUs don't become any less impressive just because they're branded "5" instead of "3".
4) While I would love the previously rumored prices to be true, that was never realistic. However, your summary here is quite naive. Small chiplets are nice, but 7nm is a far more expensive process than 14nm - rumored to be about 2x$/area. In other words, the chiplets being smaller doesn't matter much - at least not for a while. Then there's the fact that these are now MCM products with a separate I/O die, meaning both a second (relatively large) 14nm die, plus the likely quite noticeable added cost of MCM packaging. As for EOL sales making the new chips look bad? That's ... kind of how clearance sales work. "Ignore the new stuff, this is good too, and look how cheap it is!" Besides, AMD's Ryzen prices have historically always dropped in the months following launch, so there's no reason to expect this not to happen this time around.