Jobs fail if shots value comes from NumPy array

[nonhermitian]

The below code fails if the cast to int in the shots kwarg is removed:

samples = np.array([100, 250, 500, 1000, 2500, 5000, 10000])

jobs = []
for shots in samples:
    jobs.append(backend.run(trans_qc, shots=int(shots)))

The error is a bit cryptic:

Failed - TypeError: 'dict' object cannot be interpreted as an integer

[nonhermitian]

I should mention the above example used to work with the IBMQ provider.

[blakejohnson]

So, you think another reproducer would be backend.run(trans_qc, shots=np.int64(1000))?

EDIT: I have confirmed this reproducer.

[blakejohnson]

Seems like we are just missing some kind of cast in the serialization.

[drew-distefano]

@merav-aharoni please also make the update in qiskit-ibm-runtime