Fix #206 and improve Trotter–Suzuki code quality.

[cgranade]

This PR fixes #206 by adding in an appropriate /// # Remark noting the 𝑘–2𝑘 dichotomy. This PR also improves code quality in the Trotter–Suzuki decomposition feature more generally and renames DecomposeIntoTimeStepsCA → DecomposedIntoTimeStepsCA to reflect that this feature is a exposed as a function that returns a new evolution operation.

[guanghaolow]

Thanks for cleaning up the code + documentation Chris. I've added a correction on where the indexing convention differs.

[guanghaolow]

DecomposedIntoTimeStepsCA(_, 4) actually corresponds to $S_4(\lambda)$ in quant-ph/0508139.

The difference in indexing occurs in the definition of _TrotterStepSize(order: Int).

Whereas we set let stepSizeOuter = _TrotterStepSize(order); in the implementation of _TrotterArbitraryImplCA<'T>, they define the function _AlternateStepSize(order: Int) such that our stepSizeOuter is equal to _AlternateStepSize(order/2).

[cgranade]

Thanks, my apologies for misunderstanding where the index convention differed. I'll fix that, then.

[cgranade]

@guanghaolow: Thanks for double-checking and for catching the indexing error. I've expanded a lot to help prevent that confusion in the future, if you could please take another quick look. Thanks!

[cgranade]

@guanghaolow: Would you be willing to take a look at this again? Thanks for the help!

[guanghaolow]

Looks good to me!