Budget error when constant concentration or constant temperature cell is attached to other model

[mjr-deltares]

See description in issue #1529

[emorway-usgs]

Given the 3 "Constant __" packages (head, conc, temp), would it make sense to have one base class broadly referred to as "specified dv" (dependent variable)? Would boiling it down to one package help streamline the code and error handling for addressing this issue (#2311) and "retrofix" #1529?