Operations and variables

This document is meant to articulate the internal logic used for operations and variable annotations.

First note that variableAnnotationNameMap contains all variables, even those which are properties on the class. The operationVariableNameMap contains the subset of those that needs to be computed with an operation.

Second note that `operationVariableNameMap

There are a couple of important points,

1. One operation can produce many variables.
2. Variables may be cached to prevent re-computation.

When the user adds an operation, the following takes place

1. The operation is associated with its name in the operationNameMap.
2. Each variable is added to the variableAnnotationNameMap and possibly the timeDependentVariablesNameMap.
3. Each variable is associated with this operation using the operationVariableNameMap.

When the user requests a variable, the following takes place

1. Check the variable cache to see if its already been computed (if so, return).
2. Use the variableAnnotationNameMap, to find the variable annotation, and then retrieve the operation name
3. Use operationNameMap to retrieve the operation.
4. Run the operation, cache and return the results.

This seems convoluted. Why can we not just use operationVariableNameMap?

So now the user adds an operation which produces a variable that is already being used. We certainly could come up with a way to pick and choose which variables get computed. But in the short-term we need to just discard the other operation that produced the variables.