The key difference between Essentials and a new engine is that Essentials was made from scratch. A new engine would be able to look at Essentials' code and see how it does things and (potentially) base its systems off that.
Back in July/August I had a look at C# in Visual Studio for making a Pokémon game engine. That included learning C# from scratch, mind you. After about a month I had a map renderer which supported connected maps, events with multiple definable charsets (e.g. walking/running, with a definable frame rate/number per charset), autotile support (with individual frame rates), support for diagonal movement and 8-way charsets, unlimited tilesets/autotiles per map, directional tile passabilities and jumping. It also supports tiles of sizes 16x16, 32x32 and 48x48, and could auto-detect which were which.
In addition to that, I also made a map editor which was modelled on RMXP. Tile and autotile drawing works about as well as it does in RMXP, it can load and save projects, has a map tree, and so on. I've attached a screenshot of it.
If I can learn C# and Visual Studio from scratch, and come up with what I've made, as well as research deep enough into how Essentials handles maps and such in order to replicate it, then that's something. Okay, maybe I'm more talented than others, but I don't think "years and years to achieve half of what Essentials has" is accurate. Already I have a number of improvements on how RMXP and Essentials works, and it's not even at a proof of concept stage yet.