Implement BigFloat #36
Description
We require a bigfloat implementation because f64 cannot handle big scaling factors.
I am inclining towards using https://docs.rs/astro-float/latest/astro_float/ but astro_float is not compatible with BFloat trait. It should be easy to fork and implement BFloat (i.e. Num + NumOp + etc.) for astro_float. TWICE_PI can set as a static constant.
We should also implement Op<&T> for &T with Output = T type operations (like num_bigint) for BigFloat in astro_float as well. For more info check #33.