Optimise Decimal Casting

[tustvold]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

Currently the decimal conversion kernels perform unnecessary checks, this predates #6836 although is made worse by this PR.

In particular when performing a decimal conversion between the same decimal primitive (i.e. Decimal128 vs Decimal256) it should be able to elide both the checks in O::Native::from_decimal and O::validate_decimal_precision. This in turn would allow using the infallible unary kernel, likely yielding significant performance improvements.

Describe the solution you'd like

Describe alternatives you've considered

Additional context