I see two possible complications:
If you assume a general nucleus where not r_n<r_e, then the expectation value of 1/|r_n-r_e| will diverge, or at least become very large.
Secondly, even in the case of r_n<r_e, it would be difficult to calculate this integral, certainly analytically.