On the (strict) positivity of solutions of the stochastic heat equation