Add or replace a rule on a sample frame

Appends a typed rule, or replaces the rule with the same name (upsert). The payload arguments depend on type:

filter, check: column, op, values (structured) or expr (a predicate string). op in in, not_in, ==, !=, >, >=, <, <=, between, is_true, is_false.
subset: subset – a vector of entity ids to keep (captured).
require: a cross-year condition on the source, executed by resolve_frame() rather than apply_sfw(). table (default "P00"), present_in ("all" for a balanced frame, "any", or a minimum year count), and optionally a per-filing predicate as column/op/values plus holds ("every" year, the default, or "any" year). An expr string is rejected: it cannot be pushed to the source.
label: map (an id-named vector) or from (a data frame) with keys and label column names; label also names the derived column.
select: vars, scope, tables, drop (resolved via field_concordance and field_concordance_pf). scope takes a fields_in_scope() value or a variable_scope code; the 990-vs-990EZ scopes do not apply to a 990PF panel, which panelize_pf() rejects. tables takes table-name prefixes, a part code such as "P08", or a 990PF part code such as "PF01".
dedup: group, partial, amended, timestamp column overrides for deduplicate().
refresh: fn – a function taking and returning a data frame.
view: rows, cols, value, fun – a crosstab/tapply summary.
function: value/code (a string) or fn (a function).

Usage

add_rule(sfw, name = NULL, type, ...)

sfw: A sample frame.
name: Rule name. Empty/NULL auto-generates <type>_<n>. An existing name replaces that rule.
type: One of the rule types above.
...: Type-specific payload (see Details).

The updated sample frame.