stratallo

stratallo 2.2.1

Changed the name of two parameters: a to A and a0 to A0 for several functions.
Updated rnabox() so that it can perform both versions of the RNABOX algorithm, i.e.: step 1 handles one-sided upper bounds whilst step 2 handles one-sided lower bounds, or conversely, depending on the user preferences.
Minor help and vignette updates, corrected typos from previous release.

Added new user functions: ran_round() and round_oric() for rounding of (non-integer) numbers.
Renamed var_tst_si() function to var_st_tsi().
Renamed var_tst() function to var_st().
Added unit_cost variable into pop507, pop969 and pop10_mM datasets.
Renamed allocation_summary() function to asummary().
Added survey stratum cost parameter unit_costs to optcost() function.
Added survey stratum cost parameter unit_costs to rna() function.
Renamed parameter n to total_cost in rna() function.
Renamed dopt() function to opt().
Renamed nopt() function to optcost().
In nopt() function, changed the name of two parameters: D to V and b to a0.
Significantly updated help pages for most of the functions and the vignette.
Minor change in the implementation of the rna(). Vector Ri of take-bound indices is now a logical vector instead of the integer vector.
Removed h_get_which_violates() helper as it is not needed anymore
Added new argument check_violations to rna() function which replaces h_get_which_violates() approach.
Changed the parameter name from b to a0 in var_tst_si() function.
Renamed rna_onesided() function to rna().

Added new function allocation_summary() that summarizes the allocation.
Added new function rnabox() to handle box-constraints case.
Updated rna_onesided() so that it optionally returns take-Neyman, take-min, take-max set of strata indices.
Renamed function rna_one_sided() to rna_onesided().
Minor typos corrections in README, vignette.

Added new user function nopt() for minimization of total sample size under constraints on variance and upper bounds on strata sample sizes.
Re-factored rNa() into rna_one_sided() so that it also computes the allocation under optional lower bounds constraints.
Renamed nopt() do dopt() to reflect the fact that it minimizes variance function D.