It being harder to embed is definitely true, but there's increasingly will within Mozilla to change that, and if someone (e.g. MS) was willing to step up and help provide engineering resources to make it possible it's definitely doable. GeckoView is slowly changing this on mobile, and resultantly something for desktop is easier than it was five years ago.

Servo isn't replacing Gecko, though. There's nowhere near the level of resources being put into Servo for it ever to become a viable browser engine, and there's no sign that Mozilla are moving in that direction. They're still very much using Servo to experiment with new things (WebRender, parallel layout, etc.), but the approach is clearly moving modules from Servo into Gecko if/when they become production-ready. As such, Gecko isn't going anywhere, even if parts of it are getting rewritten/replaced (and that's no different to Blink!).

Servo is a research project to build a fast browser, ignoring older web standards. Project Quantum is Mozilla cleaning up those components from Servo (adding support for older standards) and migrating them into Firefox.

For refactoring a massive codebase, Project Quantum seems totally reasonable rather reasonable versus a total rewrite.

Point #1 is getting less true with GeckoView (at least on mobile), and point #2 is unlikely. Like most tech products that have a large number of users, a transition plan for Firefox would necessarily involve gradual changes (which are arguably already taking place with Quantum), which an embedder could ride.

There is no plan to EOL Gecko.