Use repeat in QuantumCircuit.power if power is numpy.integer

[t-imamichi]

Summary

Follow-up of #6831
Fixes qiskit-community/qiskit-optimization#225. Qiskit Optimization unit test of GroverOptimizer takes so long time after the merge of #6831.

This PR resolves it by applying repeat if power is np.integer.

Details and comments

GroverOptimizer of Qiskit optimization set power by numpy.random.Generator.integers. The type is numpy.int64. So, it is necessary to use np.integer for isintance to catch this case.
https://github.com/Qiskit/qiskit-optimization/blob/4953e56e3c062df42e7e51fa6b89172e1f69e3c1/qiskit_optimization/algorithms/grover_optimizer.py#L220-L232

https://numpy.org/doc/stable/reference/random/generated/numpy.random.Generator.integers.html

[jakelishman]

This was actually a mistake in #6831 - the line previously had numbers.Integral (which is what this PR should have changed it to as well), and I mistakenly changed it to int for some reason (possibly a mistake because int in type hints is equivalent to isinstance(x, numbers.Integral)). We should have the general form back, really.

[t-imamichi]

Thank you for your information. It's OK to change the code to isinstance(x, numbers.Integral).

[jakelishman]

I'm making another PR that's (hopefully) going to convert most of our isinstance(x, int) checks into isinstance(x, numbers.Integral), where that makes sense.