Prepare for rOpenSci peer review

[lbm364dl]

Adds the metadata and community files rOpenSci peer review expects, and brings
the branch back onto current main — it was 1784 commits behind, so most of
the work here was deciding what the March version still gets to keep.

What lands

File	Why
inst/CITATION	citation("whep") returned the generated DESCRIPTION default; now the package itself plus the FABIO paper
CITATION.cff	machine-readable citation, validated by cffr
codemeta.json	CodeMeta metadata; codemetar reports no complaints
.github/CODE_OF_CONDUCT.md	required for review; adopts the rOpenSci CoC
.pkgcheck	pkgcheck summary configuration
README	repostatus badge, Citation and Code of Conduct sections, link to the contributing guide

Both generated files had drifted five months: they described a package with 21
imports and three authors, against today's 25 and five.

Re-basing decisions worth a second pair of eyes

get_faostat_data() is kept. The branch's last commit removed it and the
FAOSTAT dependency as "broken". main has since gone the other way and
hardened it twice — e6922936 gave it an offline fixture so the example runs
under R CMD check, and 4ebda44b fixed the ISO3 lazy-load side channel. The
upstream API is intact too: FAOSTAT 2.4.2 is on CRAN and still exports
get_faostat_bulk() and fillCountryCode(). Re-applying the removal would have
reverted deliberate recent work, so it is reverted here instead.

main's CONTRIBUTING.md wins. The branch wrote an 86-line guide; main
has since grown a 393-line one. main's is kept as-is except its Conduct
section, which told readers to report to the maintainers while this branch's new
CODE_OF_CONDUCT.md adopts rOpenSci's, handled by their Code of Conduct
Committee. Two policies with two reporting channels is exactly what a reviewer
would flag, so CONTRIBUTING now defers to the CoC file. The rOpenSci developer
guide asks for both that and the README wording, which is theirs verbatim.

The March doc reflows are dropped. Rewrapping calculate_lmdi()'s roxygen
block split two inline code spans, and roxygen renders that as a literal space
inside the formula, so the documented decomposition string no longer parsed as
written:

emissions:activity*(activity[sector]/activity)* (emissions[sector]/activity[sector])

It also bought nothing: man/calculate_lmdi.Rd has the same four
over-80-character lines either way, because the \code{} span is emitted on one
line however the source wraps. The toy_examples.R rewrap and the any()
wrapping in test_scrape_faostat.R go with it — that line was already under 80
characters, and .bad_activity_data_param_error() returns a length-1 string, so
any() guarded nothing.

Two bugs found on the way

A DOI that does not resolve. vignettes/footprint-analysis.Rmd cited
Bruckner et al. (2019) as 10.1021/acs.est.8b03704 — absent from Crossref, 404
from doi.org. The FABIO DOI is 10.1021/acs.est.9b03554, which the same
vignette already uses correctly 550 lines earlier. Found by
urlchecker::url_check(), which rOpenSci and CRAN both run. (That pass also
reports 403 for the corrected DOI, but that is ACS refusing automated requests,
not a broken link.)

An undeclared NSE variable. R CMD check noted sovereign_iso3c as an
undefined global in .attribute_dependency_land() and
.dependency_sovereign_iso3(); it is now declared in utils::globalVariables()
as CLAUDE.md requires. This is why it stayed hidden: check-r-package@v2
defaults to error-on: warning, so a NOTE never failed CI.

On the review comments

Two of the three Copilot comments were already fixed on the branch in 2a5ed30a
(Börner's surname, in both inst/CITATION and codemeta.json).

The third asked to delete citHeader() as "deprecated since R 4.3.0". It is
not: on R 4.5.2 it is a plain wrapper around .bibentry(mheader=), and
readCitationFile() parses the file without a warning. Kept.

.pkgcheck is now honest

It suppressed four summary items; three were misreadings.

• ci_checks — pkgcheck reports "no continuous integration checks" when the
  commit it is checking has no workflow runs on GitHub yet, which is what every
  local pre-push run looks like. The package has six workflows. Unsuppressed.
• renv — renv was dropped from the package after this file was written.
  Nothing left to suppress. Unsuppressed.
• coverage — was suppressed while genuinely below the 75% threshold, hiding a
  real submission blocker. Unsuppressed, see below.
• num_imports — kept. 25 imports is what a pipeline over FAOSTAT, LUH2,
  LPJmL, CRU and HYDE costs, and pkgcheck marks it optional rather than failing.

Where this leaves the rOpenSci checklist

pkgcheck now passes on contributing file, roxygen2, URL and BugReports fields,
HTML vignette, citation file, license, and no R CMD check errors or
warnings. Two blockers remain, both needing test and documentation work rather
than metadata, so they are not in this PR:

• Coverage is 73.76%, below the required 75%.
• Ten exported functions have no examples: add_footprint_product_stage,
  add_polity_code, compute_footprint_paths, compute_fp_product_paths,
  create_alfredos_typologies, create_typologies_grafs_spain,
  create_typologies_of_josette, create_typologies_whep,
  get_polity_geometries, redistribute_feed.

Informational, not blocking: cyclomatic complexity ≥ 15 in
build_constant_territory_series (31), resolve_polity_label (27),
fill_linear (24) and build_io_model (19); and three exported names that
collide with other CRAN packages (fill_linear in transx, get_scope in FSK2R,
read_population in cape).

Separately, inst/DESCRIPTION is a tracked copy of the DESCRIPTION from around
v0.1.0.9000 — wrong version, readxl in Imports, a mipfp dependency the
package no longer has, and two missing authors. It does not affect the installed
package (R CMD INSTALL writes the real one over it, verified), so it is dead
weight rather than a bug, and removing it is left out of this PR.

Verification

• citation("whep") renders both entries and toBibtex() round-trips them.
• The FABIO entry is checked field by field against Crossref; its title now
  carries the em dash after FABIO and the en dash in Input–Output that the
  published title uses.
• The package entry takes its year from Date/Publication rather than a
  hardcoded "2025", so a release cannot silently cite the wrong year.
• cffr validates CITATION.cff; codemetar reports "the highest opinion of
  this R package".
• lintr::lint_package() clean under the project's linter set, air format .
  clean, pkgdown reference index at parity with man/ (294 topics).
• README.md regenerated from README.Rmd with github_document, not hand-edited.

Closes #592. Part of #46.

[Copilot]

Pull request overview

This PR prepares the whep R package for rOpenSci peer review by adding standard metadata files, updating the README, and creating a contributing guide. These are common requirements for rOpenSci submission.

Changes:

• Added inst/CITATION, codemeta.json, and .pkgcheck files required for rOpenSci review and proper citation support.
• Replaced the inline contributing instructions in both README.Rmd and README.md with a link to a new .github/CONTRIBUTING.md guide, and added Citation and Code of Conduct sections.
• Added a repostatus badge and excluded codemeta.json from the R package build via .Rbuildignore.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
inst/CITATION	New CITATION file with two entries: the whep package itself and the FABIO paper
codemeta.json	Auto-generated CodeMeta metadata for the package
README.Rmd	Updated with contributing link, citation section, code of conduct, and repostatus badge
README.md	Rendered version of README.Rmd with the same changes
.github/CONTRIBUTING.md	New contributing guide with code style, testing, and development workflow instructions
.pkgcheck	rOpenSci pkgcheck configuration disabling certain summary checks
.Rbuildignore	Added codemeta.json to build ignore list