Of course it is, sometimes one needs to pass data represented as R objects (like zoo) to R functions, and receiving something that is an R object and to be worked on by R calls before being passed on. That is very clumsy to manage with RCall.
Is is delightful in working with spatial raster data. It's an 'it just works' space, especially with raster and stars. There have been several attempts to disrupt the raster package but none have really stuck.
Granted, it's not fast. However trying to do the same kinds of things in python with rasterio is just cludgy. And half the time you end up making system calls to gdal anyways. Guess what? I can make the system calls from R too.