The merging principle is only when you are counting the total number of Poisson distributed events in the same time interval. But each page is a separate interval.
It is even easier: By the law of total probability, \begin{align} P(E = e) &= P(E = e \mid \text{$X$ typed the page}) P(\text{$X$ typed the page}) \\\ &\qquad\qquad+ P(E = e \mid \text{$Y$ typed the page}) P(\text{$Y$ typed the page}). \end{align} The first factor of each term is the Poisson pmf corresponding to the error rate of the typist.