surveycore v0.5.0

Breaking changes

• as_survey_replicate() replaces as_survey_repweights(). The constructor
  name now matches the underlying survey_replicate class.

• survey_nonprob and as_survey_nonprob() replace survey_calibrated and
  as_survey_calibrated(). "Calibrated" implies a post-processing step on a
  probability sample; nonprob accurately reflects the design type.

• The positional setter form set_var_label(svy, age, "label") has been
  removed. Use the named form set_var_label(svy, age = "label") instead.

• extract_var_label(), extract_question_preface(), and extract_var_note()
  now return a named character vector. extract_var_label(svy, age) now
  returns c(age = "Age in years") rather than "Age in years".

• extract_val_labels() now returns a named list. extract_val_labels(svy, sex)
  now returns list(sex = c(Male = 1L, Female = 2L)) rather than
  c(Male = 1L, Female = 2L).

• set_variable_labels(), set_value_labels(), set_question_prefaces(), and
  set_variable_notes() have been removed. Use set_var_label(),
  set_val_labels(), set_question_preface(), and set_var_note()
  respectively — all four now accept multiple variables via named ....

New features

• set_universe() and extract_universe() set and retrieve universe
  (eligibility) annotations for survey variables.

• set_missing_codes() and extract_missing_codes() set and retrieve missing
  value code vectors for survey variables.

• extract_metadata() returns all metadata fields (variable_label,
  value_labels, question_preface, note, universe, missing_codes,
  transformations) for one or more variables as a named list.

Enhancements

• All setter functions now support three call conventions: named ...
  (e.g., set_var_label(svy, age = "Age in years")), a single named
  vector/list in ..., or explicit variable = / content-argument pairs.
  All setters also now work on plain data.frames.

• All extractor functions accept multiple variables via ..., support three
  output formats ("named_vector", "list", "data_frame"), and accept a
  fill argument to include variables with no metadata in the output.