Remove `basisOpCacheVec` keyword argument

[mtfishman]

@corbett5 why is the keyword argument basisOpCacheVec needed? Can that be automatically determined from the OpSum?

[corbett5]

Sorry for the year plus delay.

One of the constraints is that there can only be a single operator acting on a given site in the OpSum, so for example a term like $Z_i Z_i$ is not allowed, instead it needs to be replaced by $I_i$. In order to automate this transformation, you can use basisOpCacheVec to provide an operator basis for each site, and then the OpSum is rewritten in this basis.

If your description of the Hamiltonian does not have multiple operators acting on a single site then specifying basisOpCacheVec is not needed.

basisOpCacheVec could also be used in the future to eliminate some of the other constraints. For example if your description of the Hamiltonian uses $I$, $Z$, and $e^{i \theta Z}$, this set of operators is not linearly independent and will therefore lead to larger MPO bond dimensions. Using basisOpCacheVec, we could expand $e^{i \theta Z}$ in terms of $I$ and $Z$. A similar thing could be done to allow the user to specify a two site operator (ie. CNOT) in the sum.