Is your feature request related to a problem or challenge?

#18032 introduced handling of decimals for pow natively; previously we'd cast the input to float before performing the pow.

It introduced a condition against non-integer & negative exponents, as the implementation can't handle them. This is technically a regression as previously we could handle them via casting to float before doing the pow.

Describe the solution you'd like

Fallback to casting value to float to perform the pow if exponent is non-integer or negative

Describe alternatives you've considered

No response

Additional context

No response