OutcomeArray handling producing unexpected numpy types with numpy 2

[ss2165]

code:

from pytket.extensions.qiskit import AerBackend
from pytket import Circuit
print(AerBackend().run_circuit(Circuit(1).measure_all(), 10).get_counts())

Outputs:
With numpy 1.26.4:

Counter({(0,): 10})

With numpy 2.1.3:

Counter({(np.uint8(0),): np.int64(10)})

[CalMacCQ]

Minor comment.

I added a workaround to the quantum phase estimation example notebook -> https://docs.quantinuum.com/tket/user-guide/examples/algorithms_and_protocols/phase_estimation.html

See the the tuple_to_str function which avoids label clashing on the x-axis of the results plot. workaround added in CQCL/pytket-docs#410

[cqc-alec]

Actually we were producing the same types in numpy 1, but never noticed because they were rendered differently.

In numpy 1, repr(numpy.uint8(1)) is 1; in numpy 2 it is np.uint8(1).