mpindex 0.3.0

Breaking changes

All argument names have dropped the . prefix (except .data). E.g. .mpi_specs → mpi_specs, .cutoff → cutoff. Passing an old dotted name now triggers an informative error naming the renamed argument and its replacement (ADR-0006).
compute_mpi() now uses a new deprived() helper for inline deprivation cutoffs. The previous .deprivation_profile list workflow is available via the new compute_mpi_from_profile().
define_deprivation() replaces the boolean .collapse argument with .collapse_fn — pass a function (e.g. max) instead of TRUE.
save_mpi() now delegates Excel formatting to the tsg package. The .formatted_output and .include_table_summary arguments have been removed.
Default .names_separator in define_mpi_specs() changed from ">" to "__".
.mpi_specs is now required in define_deprivation(), compute_mpi_from_profile(), and save_mpi(). The getOption("mpi_specs") default has been removed — pass the result of define_mpi_specs() or global_mpi_specs() explicitly.
use_global_mpi_specs() is deprecated. Use global_mpi_specs() instead.
.save_as_global_options in define_mpi_specs() is deprecated and no longer has any effect.
Minimum R version bumped to 4.1.0.

compute_mpi() — single-call API with inline deprived() cutoffs and a by argument for disaggregation.
compute_mpi_from_profile() — lower-level entry point for pre-assembled deprivation profiles (replaces the old compute_mpi() signature).
deprived() — captures a bare cutoff expression and optional per-indicator settings (collapse_fn, set_na_equal_to; .data is unchanged).
global_mpi_specs() — loads the built-in Global MPI specification. Replaces use_global_mpi_specs().
compute_mpi() returns an object of class mpi_output.
Survey-weighted MPI via weight, strata, cluster, fpc or a pre-built survey::svydesign() object (survey_design) in both compute_mpi() and compute_mpi_from_profile(). The survey package is a soft dependency (Suggests).
inference = TRUE appends design-based standard errors and confidence intervals as *_se, *_ci_low, *_ci_high columns. Confidence level configurable via ci_level (default 0.95).

Internal computation refactored to use dplyr::across() throughout, removing deprecated summarise_at() / mutate_at() / rename_all() calls.
compute_headcount_ratio_adjusted() correctly preserves ... grouping variables when aggregation is also supplied (bug fix).
tsg::rename_label() replaces manual attr(..., "label") <- assignments for column labels.

51 new unit tests for the survey-weighted path covering both input methods, point-estimate correctness, SE/CI columns, by disaggregation, and backward-compatibility.

Main vignette rewritten with plain-language explanations and a step-by-step narrative.
New vignette: Survey-Weighted MPI — covers both input paths, inference columns, subgroup disaggregation, and a weighted vs. unweighted comparison.

Changed returned value of define_mpi_specs from list to data frame with a class of mpi_specs_df. By default, it stores a global option named mpi_specs which can be accessed using getOption('mpi_specs'). Other specifications are now stored as attributes.
Changed default value of .unit_of_analysis from households to NULL.
save_mpi now allows unformatted output (tidy format) when generating an output.
Implemented feature to include MPI specification in the output by setting .include_specs tot TRUE in save_mpi.
Deprecated .names_separator argument in define_mpi_specs.
Enhancement of define_deprivation based on #17.
Added use_global_mpi_specs() wrapper function to use Global MPI specification.