Numpy dtype handling?

[Nomos11]

Many isinstance-checks check for float or int. This fails when numpy-dtypes like float64 are used. So far this does not induce errors, but for future considerations that might be worth keepig in mind.

[shumpohl]

This depends on the nature of the isinstance checks, I think. These are preliminary thoughts:

In cases where a numeric value is expected as an input we should try to handle all Real values, that is integers, floats and fractions of all kinds (builtin, numpy, sympy, gmpy2, ...). Ideally, not with isinstance checks.

But in cases where we can have arbitray objects (nested structures, dicts, str) we should try to enforce the caller to provide standardized numeric types.

We have qupulse.expressions.sympy.ALLOWED_NUMERIC_SCALAR_TYPES as a hack right now.