Additional issues on validator

[cboettig]

From @mdietze:

Some additional thoughts I wanted to jot down:

• I want to start outlining the standard in a Rmd file in this repo so that it's easy to knit to different file formats
• should make sure the output variables are a required part of the metadata
• would be good to have an optional way to map IC and outputs (are there examples were the IC variables are not just a subset of the output variables?)
• Need to have a way of specifying whether process error contains covariances
• Might also add a way of distinguishing sampling/demographic stochasticity from other process errors as Raiho found DS inadequate to capture process error, but I could see a lot of folks with DS checking off "yeah, I've got process error in my model"
• Would be good to have an optional method for documenting parameters, etc not just recording their numbers
• Need to follow up on previous discussion about whether/how to augment output file with parameters used by ensemble members
• ...