Need to have dummy/fake/place holder observer value in a MRDS dsm for dsm to work

[TiagoAMarques]

For some reason the code in dsm requires a value for observer, even though it does not use the variable in the analysis. This makes no sense as dsm uses only one row per observation, and hence, for duplicates, one cannot define a single observer. A current workaround it to create a variable with NA's in it to make the code work, but it might be worth recoding dsm to avoid this nonsensical step.