Wire up peace-fwcp.Rmd interpretation citations from 3-split lit reviews

[NewGraphEnvironment]

Problem

The peace-fwcp.Rmd vignette interpretation paragraphs (the
non-snow ones) make claims about temperature, precipitation, and
drying that currently land on zero peer-reviewed citations.
The 3-split climate-departure lit reviews (#53/v0.1.7,
#58/v0.2.2, #61/v0.2.3, #63/v0.2.4) plus the snow lit review
built the citation backbone; the kootenay-lake vignette was wired
up in #65/v0.2.5. Now the peace-fwcp vignette gets the same
treatment.

Scope

peace-fwcp.Rmd only. Mirrors the #65 (kootenay-lake) wire-up
pattern verbatim. Per feedback_vignette_citations_sparse.md:
cite sparingly, only where finding is visible in graphs/tables,
plain language, no decoration.

Pull [@key] markers from the 4 cite-this-for-that maps in:

• planning/archive/2026-05-issue-53-snow-lit-review/findings.md
• planning/archive/2026-05-issue-58-temperature-lit-review/findings.md
• planning/archive/2026-05-issue-61-precip-drying-lit-review/findings.md
• planning/archive/2026-05-issue-63-interpretation-framing-lit-review/findings.md

Per-AOI differences vs Kootenay

The Peace AOI (~73,000 km², 16 WSGs across 5 ecoregions) has
different climate-departure findings than Kootenay:

• Precipitation up in 2 of 5 ecoregions (BMP, NRM); flat
  regional mean; this is where Peace and Kootenay diverge
  sharpest
• Freshet timing shift uniform across all 5 ecoregions
  (~1 day/decade earlier melt)
• Day-night asymmetry subtler than the textbook signal —
  may or may not warrant Karl 1993 cite depending on whether
  it's visible in DTR plot (Kootenay's was)

Cite candidates likely overlap heavily with #65 (Hansen, Arguez &
Vose, Karl, Pepin/Rangwala, Mantua/Eaton) — but Kang 2016 may
be load-bearing for the Peace since Kang explicitly studied the
Fraser-and-Peace freshet system. May also need additional cites
specific to Peace findings (e.g., precipitation-up signal —
already in collection refs).

Audit trail (load-bearing, same as #65)

For every [@key] added, capture in
planning/active/citation_audit.md:

1. Vignette excerpt where the cite is added
2. Source quote/paraphrase
3. Rag store + topic where retrieved
4. Paraphrase as written in the vignette
5. Why warranted (visible in graphs/tables)

Independent verification

Spawn an Explore review agent on the audit log (same pattern as
#65). Address feedback before PR.

Out of scope

• kootenay-lake.Rmd — done in Wire up kootenay-lake.Rmd interpretation citations from 3-split lit reviews #65/v0.2.5
• New lit reviews — 3-split + snow are the source corpus
• Vignette section restructuring — only inserting [@key]
  markers + minor prose smoothing
• FWCP Peace cross-references in kootenay-lake.Rmd (L528,
  L973, L996) — separate small follow-up not addressed in
  either Wire up kootenay-lake.Rmd interpretation citations from 3-split lit reviews #65 or here

Done when

• peace-fwcp.Rmd interp paragraphs cite authorities for
  findings visible in graphs/tables, sparingly per philosophy
• planning/active/citation_audit.md has one row per [@key]
  added with the 5 fields
• Independent review agent ran on audit log; flagged issues
  addressed
• Vignette renders cleanly with all [@key] markers
  resolving
• No vignette section restructuring; only citation insertion +
  minor prose smoothing
• PR + release v0.2.6 (patch — vignette docs)