When were the standard bump function examples such as $e^{-1/(1-x^2)}$ first understood, and what was the context or motivation at the time?
As an upper bound I would guess that they must have been understood by the time of Émile Borel's 1895 Sur quelques points de la théorie des fonctions, since Hörmander cites that article as proving Borel's lemma. However, I am unable to read French to see if Borel introduces it or is developing prior work.