Optionally use QD or DD library for longdouble
Issue #17
new
At the moment longdouble is either __float128 or long double. Would be nice to support some other long doubles.
QD library might be required in temponest somewhere, but not sure.