ARSAU dataset loaders + registry (R-side mirror of morie.arsau_datasets)

ARSAU = the Ontario Ministry of the Solicitor General's provincial release of Police Use-of-Force incident records (formally "Race-Based and Identity-Based Data on Police Use of Force in Ontario"). Published on the Ontario Data Catalogue at https://data.ontario.ca/dataset/police-use-of-force-race-based-data.

Details

This file ships the R-side equivalents of the Python morie.arsau_datasets module:

• ARSAU_REGISTRY(): returns the registered (year x kind) entries as a list of lists.

• morie_arsau_load_main_records(), morie_arsau_load_individual_records(), morie_arsau_load_probe_cycle_records(), morie_arsau_load_weapon_records(), morie_arsau_load_aggregate_summary(), morie_arsau_load_detailed_dataset(): per-record-type loaders, returning a named list with data, schema, sidecar, year, kind, language, is_valid, n_rows, n_cols, interpretation.

• morie_arsau_available_years(), morie_arsau_available_datasets(), morie_arsau_describe(): discovery callables.

Path portability

No path on the maintainer's workstation is hard-coded. All file resolution goes through .morie_resolve_arsau_dir (defined below), which honours, in order:

1. an explicit data_dir = argument

2. the MORIE_ARSAU_DIR environment variable

3. MORIE_DATA_DIR/arsau

4. morie_cache_dir("arsau") (only if already populated by a previous morie_arsau_download() call – never auto-created at read-time, per CRAN policy)

5. system.file("extdata", "arsau", package = "rmorie") – the bundled tiny fixture for unit tests + tutorials

6. stop with a remediation paragraph

2023 weapon-records invalidity gate

The 2023 release ships uof_weapon_records_invaliddata.csv, flagged by the ministry as non-compliant. morie_arsau_load_weapon_records(2023) signals an error unless the caller passes allow_invalid = TRUE; when allowed, the returned object's is_valid field is FALSE and its warnings list opens with an explicit caveat paragraph.