API helpers for non-finite float validation

[NailxSharipov]

Extend API for finite/non-finite validation.

• FloatNumber

  • is_finite()

• FloatPointCompatible

  • is_finite() with default impl

Use it in:

• FloatPointAdapter::validate_scale
• FloatPointAdapter::try_float_to_int

as bonus fix: src/adapter.rs
fn validate_scale(scale: P::Scalar) -> Result<f64, FloatPointAdapterScaleError>

using new template api

[LucaCappelletti94]

I believe that for floats we should just use https://docs.rs/num-traits/latest/num_traits/ which already comes with the described APIs.

May I ask why you redefined the Float trait instead?