Fix method ambiguity error of NaNMath.pow

[devmotion]

NaNMath 1.0.2 introduced a method ambiguity error.

[devmotion]

Hmm seems not sufficient...

[AayushSabharwal]

Suggested change

	NaNMath.pow(x::BasicSymbolic{<:Number}, y::Integer) = x^y
	NaNMath.pow(x::BasicSymbolic{<:Number}, y::Integer) = maketerm(typeof(x), NaNMath.pow, [x, y], metadata(x))

[devmotion]

We want to avoid NaNMath in that case in NaNMath < 1.0.2 and on NaNMath >= 1.0.2 an Integer as second argument falls back to ^ anyway. So I think using ^ is the better and safer approach, regardless of the NaNMath version.

[AayushSabharwal]

Then can we add this to @number_methods?

[devmotion]

I did but I guess incorrectly. The original error is gone but it seems it causes https://github.com/JuliaSymbolics/SymbolicUtils.jl/actions/runs/12708708876/job/35426212172?pr=690#step:6:885.

[AayushSabharwal]

In fact, this should instead be a change to @number_methods which adds a NaNMath.pow(::T, ::Integer)

[AayushSabharwal]

Actually, never mind 😅 Symbolics is the only one with the ambiguity, right? Because Num <: Number. Adding a method here doesn't help. We instead need NaNMath.pow(x::Num, y::Integer) = wrap(NaNMath.pow(unwrap(x), y)) in Symbolics

[AayushSabharwal]

See JuliaSymbolics/Symbolics.jl#1398

[AayushSabharwal]

I think a couple of the tests need to be updated