Avoid overflow in passing large exponents to singular

[user202729]

Fixes #40442

I don't know what the output would be in 32-bit.

I think it's justified to change singular_polynomial_pow's exp parameter type from unsigned long to int, because previously it didn't work for int anyway (since singular's pPower takes int parameter). We could keep making it take unsigned long and do the divmod INT_MAX at low level, but I find it easier to do that at high level (Python), where, as can be seen from this pull request, that takes only 3 lines of code.

I think it's completely impractical to raise anything except monomials to power larger than 2^31 anyway. Except maybe (x+y)^2^n mod 2, but that can only be quickly calculated when internally repeated squaring is used.

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies