i regularly use K (an APL descendant) as "pseudo-code".
in fact, i regularly _think_ about problems in an array-programming fashion, heavily inspired by K. when jotting them down, i often use K. when turning them into actual code to run, it's mostly immaterial if i then use K, J, numpy, julia, etc. the general approach translates well into a multitude of array programming varieties.
as such, i don't believe your point 5 holds. this way of thinking is certainly off mainstream, as is, e.g constraint programming, but there are niches where people use it highly productively. you won't necessarily glean that from the output coming from those niches, though.
in fact, i regularly _think_ about problems in an array-programming fashion, heavily inspired by K. when jotting them down, i often use K. when turning them into actual code to run, it's mostly immaterial if i then use K, J, numpy, julia, etc. the general approach translates well into a multitude of array programming varieties.
as such, i don't believe your point 5 holds. this way of thinking is certainly off mainstream, as is, e.g constraint programming, but there are niches where people use it highly productively. you won't necessarily glean that from the output coming from those niches, though.