Add join_smq() to expand SMQ lists into per-LLT dictionary rows

[billdenney]

Fixes #14. Adds an exported join_smq() that turns the SMQ definitions parsed by read_meddra() into the flat per-LLT term list needed to merge against coded adverse events and to build SMQ*/CQ* variables.

Summary

Fixes #14. Adds an exported join_smq() that turns the SMQ definitions
parsed by read_meddra() (smq_list.asc + smq_content.asc) into the flat
per-LLT term list needed to merge against coded adverse events and to build
SMQ*/CQ* variables.

join_smq(data) returns one row per (SMQ × LLT × SOC path) with the full
SOC → HLGT → HLT → PT → LLT hierarchy plus SMQ identity and member
attributes (term_scope, term_category, term_weight, term_status).

How it works

• Sub-SMQ recursion — term_level == 0 rows reference a child SMQ and are
  flattened recursively via a cycle-safe worklist.
• PT → LLT expansion — term_level == 4 PT members expand to all LLTs of
  the PT, with scope/category inherited.
• Direct LLT members — term_level == 5 attach to the single LLT.

It reuses join_meddra() for the hierarchy and uses inner_join, so orphan
terms with no hierarchy match are dropped rather than emitting all-NA rows
(the one intentional departure from join_meddra()'s tolerant left_joins).

Output columns use the same snake_case names as join_meddra(). Per the
agreed design, the vignette documents a recipe to rename to the SAS/ADaM short
names from the issue (soccd, lltcd, ptcd, …).

Changes

• New: R/join_smq.R (join_smq() + internal flatten_smq_content() /
  flatten_smq_one()), tests/testthat/test-join_smq.R, man/join_smq.Rd.
• Updated: NAMESPACE (export), NEWS.md, vignettes/meddra-read.Rmd
  (new "Expanding an SMQ into dictionary terms" section + rename table),
  inst/WORDLIST, CLAUDE.md.

Testing

• devtools::test() — all pass (7 new tests: PT, LLT, sub-SMQ, cycle-safety,
  missing-file, empty-content, empty-list paths).
• Coverage — 100% of R/join_smq.R (covr).
• devtools::check() — 0 errors, 0 warnings; only environmental NOTEs
  (.git present; clock check).
  join_smq() returns one row per (SMQ x LLT x SOC path) with the full SOC -> HLGT -> HLT -> PT -> LLT hierarchy plus SMQ identity and member attributes (term_scope, term_category, term_weight, term_status). It:

• recursively resolves sub-SMQ references (term_level 0, cycle-safe);
• expands PT members (term_level 4) down to all of the PT's LLTs, with scope inherited; and
• attaches LLT members (term_level 5) directly.

It reuses join_meddra() for the hierarchy and uses inner_join so orphan terms with no hierarchy match are dropped rather than emitting all-NA rows. Column names match join_meddra() (snake_case); the vignette adds a recipe for renaming to the SAS/ADaM short names.

Also: regenerated NAMESPACE + man/join_smq.Rd, extended the vignette SMQ section, updated NEWS.md, CLAUDE.md, and inst/WORDLIST. Tests cover PT, LLT, sub-SMQ, cycle-safety, missing-file, and empty-input paths (100% coverage of R/join_smq.R); R CMD check passes with no new NOTEs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
R/join_smq.R	100.00% <100.00%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.