This should fix issue #1015. I have reimplemented str_to_float to extend the grammar that it can parse. Including, say, strings with a decimal part :)

I take the opportunity to replace the implementation of int::pow to use fast exponentiation, to add a few functions to the new float module, as well as a few tests. I also renamed str_to_float and float_to_str into float::from_str and float::to_str, to harmonize with the int module.