Contribution identifiers

[bfoley12]

Summary

Modified how Contributions are identified.

Previously:

• Uniqueness was only sometimes guaranteed by {Contribution.project, Contribution.identifier} (identifier is a user-defined field - it can be random, but was often set to an mp-id) - only true when it's Project had Project.unique_identifiers == True
• If Project.unique_identifiers == False, the only way to get a unique Contribution was by querying the Contribution._id, which is an ObjectId - very cumbersome

This branch:

• Removed Contribution.identifier
• Contributions are guaranteed to be identified uniquely by {project, material_id, chemical_system_id, formula, unique_value?, condition_key?}:
  • Contribution.project: unchanged, the project the contribution belongs to
  • Contribution.material_id: the mp-id of the material being studied
  • Contribution.chemical_system_id: the chemical system string
  • Contribution.formula: unchanged, the formula of the system
  • Contribution.unique_value: optional, but required in Contributions that are not uniquely identified by the previous fields. This is specified by the user as a Contribution.data field, then extracted by the server into unique_value, so we can define a MongoDB index on it
  • Contribution.condition_key: Added in anticipation of another branch that 'pivots' contributions when conditions are present. Not fully developed, but set as a harness to use later.
• Added validation to identifier fields
  • material_id: must start with "mp-" and be followed by numbers or letters (matching MP- and AlphaId formats)
  • chemical_system_id: hyphen-delimited list of valid Elements
  • formula: NFKC normalized string of valid elements and numbers (trimming leading 0s)
• Replaced Project.unique_identifiers with Project.unique_column to signal which Contribution.data field is the Contribution.unique_value.