feat: cog_categories

Discovery verb over the summary_categories view, grouped one row per
(category, subtype). Parallels cog_gov_search: analysts use it to
find the valid `category` values to pass into cog_spending(),
cog_revenue(), cog_geographic_rollup().

Columns: category, category_type, subtype, n_codes, item_codes
(comma-separated, alphabetical). Optional filters:

  type    = NULL | "spending" | "revenue"
  pattern = regex matched case-insensitively on category

The user-facing "spending" alias is translated internally to the
corpus-native "expenditure" so callers don't have to learn Census
vocabulary, while the returned category_type column preserves the
native value for auditability.

Also: fix @noRd placement in session.R so devtools::document() stops
warning.

Tests: +16 new / 181 total pass. check 0E/0W/0N.

NAMESPACE

@@ -1,5 +1,6 @@
 # Generated by roxygen2: do not edit by hand
 
+export(cog_categories)
 export(cog_explain)
 export(cog_find_peers)
 export(cog_geographic_rollup)