Well, things are never that simple 
of course the irony is that the overarching concern expressed in the thread I linked to is about compiler toolchains (best summarised in Is adding new Fortran to SciPy generally allowed? - #9 by rgommers - scipy - Scientific Python and Is adding new Fortran to SciPy generally allowed? - #11 by rgommers - scipy - Scientific Python), and LFortran could be a key ‘missing piece’ in that respect.
We just find ourselves in a bit of a chicken-or-the-egg situation: it is hard for LFortran to commit to SciPy while the prospect of removing all Fortran looms, and it is hard for SciPy to commit to Fortran given the uncertain future of compiler toolchains (along with a few other concerns described in The difficulties of using Fortran in SciPy - #54 by ev-br).