build_univariate_ord_problems: Build a list of univariate ordinal problems
In MichaelHoltonPrice/yada: Yet Another Demographic Analysis package

The data directory (data_dir) contains a main problem file uniquely determined by the analysis name (analysis_name), and possibly some corresponding fold problems. Return a list of univariate problems to solve. The model specifications to include in the returned list are specified by mean_specs and noise_specs, which must be the same lengths. By default, all combinations of mean specifications and noise specifications supported by yada are used; there are three ordinal mean specifications (pow_law_ord, log_ord, and lin_ord) and two noise specifications (const and lin_pos_int), so by default there are num_models = 6 six total model specifications. By default, cross validation fold problems are not included, in which case the return list, ord_prob_list, has length num_models * J. If the cross validation problems are included, ord_prob_list has length num_models * J * (1+num_folds).

build_univariate_ord_problems(
  data_dir,
  analysis_name,
  mean_specs = c(rep("pow_law_ord", 2), rep("log_ord", 2), rep("lin_ord", 2)),
  noise_specs = rep(c("const", "lin_pos_int"), 3),
  add_folds = FALSE
)

`data_dir`	The data directory with problems and results
`analysis_name`	A unique analysis name (for the input data directory)
`mean_specs`	A vector of mean specifications for the ordinal models (default: [c('pow_law_ord','pow_law_ord','log_ord','log_ord','lin_ord','lin_ord'])
`noise_specs`	A vector of noise specifications for the ordinal models (default: [c('const','lin_pos_int','const','lin_pos_int','const','lin_pos_int'])
`add_folds`	Whether or not to include cross validation folds in the return list