Clean up descriptor sets to make lesser-adopted sets more semantically consistent

Context

In the Ed-Fi UDM descriptor sets are always mutable - the set can be modified to support local context and needs. Long experience has shown that there is no value set that can meet all use cases for the same concept.

However, this mutability applies only to the UDM: more narrow use-case-focused standards – such as the Ed-Fi REST API standards –  do prescribe usage of specific values for more narrow use-case focused standards.

Nevertheless, analysis of field usage has revealed that some descriptor sets are highly mutable and unlikely to be standardized - the values tend to vary extremely. A good example is:

StaffEducationOrganizationContactAssociation.ContactType 

This could be anything from "CRDC Coordinator" to "Grade-level Coordinator" to "Math Chair" to "Volleyball Coach" - values are theoretically infinite, and at a UDM level are not ever likely to be standardized.

"local" values can be found here: https://techdocs.ed-fi.org/display/EFDS32/Unifying+Data+Model+-+v3.2+Model+Reference 

Example page: https://techdocs.ed-fi.org/display/EFDS32/Assessment+Domain+-+Entities%2C+References%2C+and+Descriptors 

Task

For descriptors that are in this bucket - i.e. nearly always highly localized – the task is to make the "default" values more semantically consistent. This means:

• coordinate such values around a coherent set of defaults
• remove outlier values (many of these sets have accumulated values which are clearly implementation-specific and rarely used)
• look for values that don't follow Ed-Fi conventions (e.g., use acronyms, use codes as opposed to expressive values)