Person `Dcpst` enum still returns 'PreviouslyPartnered' which isn't generated in initial populations

[andrewbaxter439]

Problem

• Merging in multiple pull requests #294 merges in changes discussed in Data&estimates update 5nov #283 which makes dcpst (partnership) into a two-category variable: partnered and single
• regression estimates derived from initial population files generate one coefficient (e.g. Dcpst_Single over baseline 'partnered')
• getDcpst still returns PreviouslyPartnered when should return Single for newly singly persons
• thus the Dcpst_single coefficient is estimated for this population but never applied

https://github.com/centreformicrosimulation/SimPaths/blob/bb58b85989cdd377477e7c590b3630793971fba1/src/main/java/simpaths/model/Person.java#L4648-L4659

Suggestions

Either:

• if 'previously partnered' is not intended for use anymore anywhere then remove from getDcpst and return 'single' when relevant
• or estimate the three-category effects on a separate dummy to restore 'previously partnered' effects.