I've been wondering about this. On one hand, Godot might not want to rely on external developers for such a core feature of the engine, they also might want to/plan on implementing it in a completely different way. The author of Terrain3D might not even be interested in collaborating on that level and be satisfied with it remaining a GDExtension, for all we know.
However, I think it would be a very good thing for various reasons. First off, Godot needs this feature eventually, it's very important. If they developed an alternate core terrain system, it might discourage the creator of Terrain3D to continue work on this already capable system. If they fully integrated it, support it, and continue to improve on it, then no work is wasted, and it incentivizes people to create more very ambitious extensions for Godot in the future, without fear that their work might eventually be in vain, deprecated, or made obsolete.
We can see this play out in the .NET ecosystem when popular Nuget packages exist and are used by everyone, they are so popular and core to the way of doing things that .NET eventually creates their own version of it, usually improving on it, which on a community level kills off interest in the existing package. A lot of developers see this as very problematic and unhealthy to the .NET ecosystem.
Either way, I haven't seen any word on this from Godot developers or Terrain3D developers, I'm curious if anything has been said on the matter.
Tl Dr; I think integrating Terrain3D would increase people's faith in relying on GDExtensions and increase people's confidence in developing ambitious GDExtensions. If Godot has a better idea on how to implement it, then I think it would be better to work that implementation into Terrain3D as it exists right now.