Operator converts tuple to np.ndarray

[albi3ro]

Currently, Operator converts tuples to np.ndarray upon initialization:

>>> op = qml.BasisState([1,0], wires=(0,1))
>>> op.data
[array([1, 0])]

But it does not do the same thing with tuples. Somehow previously, operators like BasisState could accept tuples as their parameters, and at some point (we think in Unwrap), they got converted to the expected np.ndarray. Since this no longer happens in Unwrap, we now explicitly convert tuples to numpy arrays as well.

Lightning tests used tuples, and thus expected this behaviour to still exist. Instead of changing those tests, I think it makes to continue accept tuples in addition to just lists.

Note: I moved some things around in test_operation.py so the tests would be a little bit better organized.

[mlxd]

Thanks for the quick turnaround @albi3ro
For non-trainable parametric gates like BasisState I think the old behaviour that this returns to makes sense.

[timmysilv]

looks good!!

[codecov]

Codecov Report

Merging #4022 (03793c0) into master (b9071bc) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #4022   +/-   ##
=======================================
  Coverage   99.58%   99.58%           
=======================================
  Files         351      351           
  Lines       30847    30847           
=======================================
  Hits        30718    30718           
  Misses        129      129           

Impacted Files	Coverage Δ
pennylane/operation.py	97.28% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.