Fix spin_op multiplication.

[boschmitt]

Description

The spin operator class, spin_op, gives wrong results when multiplying. The following output demonstrates that how it gets wrong the multiplication of single-qubit, single-term spin operators:

**********************************************************
Testing 1 qubits
**********************************************************
I * I = I == I [OK]
I * X = X == X [OK]
I * Y = Y == Y [OK]
I * Z = Z == Z [OK]
X * I = X == X [OK]
X * X = I == I [OK]
X * Y = iZ == iZ [OK]
X * Z = -iY == -iY [OK]
Y * I = Y == Y [OK]
Y * X = -iZ != iZ [FAILURE]
  [Coefficient error] got: (0,1) expected: (0,-1)
  Sanity check:
    [q0] Y * X = -iZ

    Expression: (-iZ) = -iZ
Y * Y = I == I [OK]
Y * Z = iX == iX [OK]
Z * I = Z == Z [OK]
Z * X = iY != -iY [FAILURE]
  [Coefficient error] got: (0,-1) expected: (0,1)
  Sanity check:
    [q0] Z * X = iY

    Expression: (iY) = iY
Z * Y = -iX != iX [FAILURE]
  [Coefficient error] got: (0,1) expected: (0,-1)
  Sanity check:
    [q0] Z * Y = -iX

    Expression: (-iX) = -iX
Z * Z = I == I [OK]

This PR fixes the issue, adds more comprehensive testing, and simplify the logic behind doing the cartesian product of the terms of two spin_op objects, which shows better performance on some tests (YMMV).

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[boschmitt]

/ok to test

Command Bot: Processing...

[boschmitt]

/ok to test

Command Bot: Processing...

[sacpis]

Thanks @boschmitt. LGTM!

[amccaskey]

/ok to test

Command Bot: Processing...