Fixed `dace::math::pow` for Integer case

[philip-paul-mueller]

Added a pow specialization for dace::math::pow.

The new specialization works for integers if all arguments are integer the return value is guaranteed to be integer as well. This should also solve the issue with the fft test mentioned by Tal (https://github.com/spcl/dace/pull/1742/files#diff-a81c4594b19da98f6516283a50f6958b04976ae2913f097358141b2681868f3eR158). However, to work we need https://github.com/spcl/dace/pull/1747.

[tbennun]

@philip-paul-mueller That's a lot of modern C++. Can you verify that it conforms to C++14?

[philip-paul-mueller]

Yes everything is C++14, actually it looks a bit strange because for example std::is_integral_v is in C++17. We should consider switching to C++17.

I spotted a small error in the current implementation and also one in ipow which I fixed.

[philip-paul-mueller]

Actually If I have time I want to go through the runtime and fix some things there (for example the range class in pyinterop.h is wrong if (end - begin) % skip != 0 holds.

[tbennun]

Yes everything is C++14, actually it looks a bit strange because for example std::is_integral_v is in C++17. We should consider switching to C++17.

I spotted a small error in the current implementation and also one in ipow which I fixed.

We cannot do so yet, definitely not before 1.0. As far as I understand some of our target platforms do not support it and it will change our compiler requirements.

[philip-paul-mueller]

I currently have a strange error with template instantiation. I am looking into it as soon as I can.

[philip-paul-mueller]

@tbennun On the note about C++17, in HIP mode C++17 is used. https://github.com/spcl/dace/blob/02c9c3787d0c16c5a7a8ba006bf446d24c9a0443/dace/config_schema.yml#L313