WP 34S and 31S bugs and fixes

[emece67]

I've found what I think is a bug.

In short, the complex version of y^x gives a domain error, instead of 0+0j, when the argument in ZT is also 0+0j.

This happens with any exponent in XY.

Looking in the sources seems that the complex y^x operation is computed via exp(ln(y)*x), but the code does not check the special case y = 0 prior to computing ln(y). The complex cubic root also suffers from this.

Regards.