Added template overloads for integer_times_pow10()

[toughengineer]

Since the question was raised in #320, I added template overloads for integer_times_pow10().

Usage:

const auto a = fast_float::integer_times_pow10(23, 42); // a is a double
const auto d = fast_float::integer_times_pow10<double>(23, 42); // d is a double
const auto f = fast_float::integer_times_pow10<float>(23, 42); // f is a float

Things to pay attention to:

• please see comments in files.

Will add some docs to README later.

[toughengineer]

This is kinda ugly, an option is to add a macro similar to FASTFLOAT_ENABLE_IF along the lines of

#define FASTFLOAT_ENABLE_IF_T(T, ...) typename std::enable_if<(__VA_ARGS__), T>::type

Then even the macro FASTFLOAT_ENABLE_IF can be reimplemented (DRY and whatnot):

#define FASTFLOAT_ENABLE_IF(...) FASTFLOAT_ENABLE_IF_T(int, __VA_ARGS__)

[toughengineer]

I don't understand this, but, apparently, to really overflow FLT_MAX, +0.00000001e38 (i.e. seven zeros and one after the decimal point) is not enough.
I verified this with a different implementation and it seems to be correct.