Naming coincidence, they are in fact unrelated.

A `monad` in Category Theory (which is where Haskell gets them from, ultimately) is a single-argument functor and two natural transformations, so there probably is a through-line here in terms of one-argument function-likes.

APL's use of 'monad' predates Haskell's existence by decades.

...and Leibniz's "windowless monads" (all three concepts are pairwise distinct, related solely by the greek prefix for 'single') predates APL and Haskell by centuries.