BUG: Non-weak promotion of Python integers can be cumbersome for unsigned integers

[seberg]

This is to gather and close some other issues. The NEP 50 proposal would close it. In many cases using int64 as the default Python integer for scalar operations (where we disable value-based logic) can be surprising

• bitwise_and with uint64 #2955 (bitwise_and)
• Cannot reliably compare uint64 values to integers? #5746 (comparison) although uint64 and int64 comparisons of course have still more issues (covered by BUG: incorrect equality of np.int64 and np.uint64 #12525)
• np.left_shift and np.right_shift fail on np.uint64 scalar types (Trac #1931) #2524 (bit shift operators)
• BUG: Inconsistent behaviour of return type of int64.__pow__ #8809 power is surprising for promotion (which is just the normal madness of scalar+value-based logic), again together with BUG: incorrect equality of np.int64 and np.uint64 #12525
• uint64 converted to float when dividing by python integer #13057 (division of uint by Python int)

This issue is just to gather and close the other ones as duplicates.