Note the motivation in that Intel forum thread is the same as in Modern Fortran Explained, by Metcalf et al., 7th Edition published back in year 2011, nearly a decade ago. In section 20.14.3 Denoting absent arguments, the authors offer what they themselves term as a somewhat contrived situation involving several optional dummy arguments that the consumer may want to use with ALLOCATABLE or POINTER variables only if allocated or associated which can then lead to a combinatorial explosion of IF constructs without this facility.
Language purists are likely to scoff at such features, especially the one with ALLOCATABLE objects that can indeed come across as orthogonal to the notion in the language of not referencing such an object when unallocated. But these features are otherwise rather useful for the practitioners of Fortran.
My view is the long lag time with compiler implementations of such miscellaneous “goodies” adversely impacts awareness and adoption, cue the reactions upthread. If these facilities can be seen instead as “low-hanging fruit” by compiler developers and implemented with eagerness at short notice, that can be a boon for Fortranners.