I've applied a fix that fixes the two bug listed above, but it was a
bit of a hack. The essence of the fix was to make a form that was
easier for maxima to factor.
Barton comments in Bug 938235 that he added a call to gfactor to
solvecases and also add an merror to polelist.
What should we do about these? For these particular integrals, the
problem happens in zmtorat and the call to $factor. The integrand has
been transformed and zmtorat is now trying to find the factor the
denominator to find the poles. $factor fails, so zmtorat gets
confused.
At the very least, it seems an merror for polelist is a good idea.
Ray