The purpose of using space-filling curves is to convert coordinates into 1 dimensional data, thus they can be indexed by ordinary database index, no matter distrusted or not. The problem is it's much slower than a "real" spatial index like R-tree, and the supported queries are rather limited.

On the other hand, spatial data can be easily partitioned, rendering the aforementioned method less useful. Also, I've never seen anyone needed more than a single instance of postgres to index all the spatial data, unless you are dealing with spatial temporal data, which can grow infinitely large. Then again, spatial temporal data can be easily partitioned.