I'm guessing pragmatism and practicality is why they went with Elm over PureScript? If their backend is in Haskell then it seems like they have devs who know it well enough to use PS.
When you hit a wall you can probably work around it in Purescript whereas I feel you are too sandboxed in with Elm and you'd have to start patching your Elm source to get around it.
I actually find PureScript quite practical by design and goals (for example the ffi and readable generated code).