Two improvements in one pass:

1. Tightened maps. All context layers (towns, lakes, rivers, streams,
   highways) are now clipped to the AOI watershed group via st_intersection
   in the load-context chunk; the two spatial-departure rasters are masked
   to the AOI so cells outside the watershed render transparent. Replaces
   the previous bbox-based view that showed features extending past the
   WSG boundary.

2. New "Day-Night Asymmetry" section between Seasonal Patterns and
   Precipitation/Soil Moisture. Three figures showcase the tmax/tmin
   data now on STAC:
   - tmax annual anomaly time series
   - tmin annual anomaly time series
   - diurnal range (tmax - tmin) trend

   Ecological framing covers ET intensification from warmer nights,
   weakened cold-air pooling and pest-overwintering implications, and
   the stream thermal regime consequence for salmon-bearing watersheds.

Code-checked. Two minor doc-quality fixes from review applied:
- Per-layer st_agr("constant") instead of broad suppressWarnings()
  so genuine geometry warnings still surface
- Added fig.cap to plot-tmax and plot-tmin so they're cross-referenceable

Vignette renders cleanly; HTML size 2.69 MB.

Relates to #39

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Iteration after looking at the actual KOTL trend data:

- Narrative now matches the data. The textbook day-night asymmetry
  does not show at this watershed (diurnal range trend is -0.06 C
  over 75 years, well within noise). Pivoted the section to focus
  on the strongest signal that DOES show: summer daytime maximum
  has warmed 2.8 C since 1951.
- Interpretation rewritten. Previous text claimed precipitation
  trends are weak — the data show -10.7% over 75 years at
  p = 0.015, a real significant decline. Soils are drying because
  precipitation is down AND warmer temperatures drive higher
  evapotranspiration; both compound.
- Salmonid not salmon, with explicit note that anadromous runs to
  Kootenay Lake are blocked by lower-Columbia dams. Resident
  salmonids and First Nations reintroduction efforts are the
  ecological hook for summer thermal stress.

Other improvements:
- cd_compare now demonstrated with method="pct_change" for
  precipitation and soil moisture, alongside the absolute mean_diff.
- kableExtra scroll_box wraps the long tables (catalog, baseline,
  compare, all-trends) so they don't dominate vertical layout.
  Added kableExtra to DESCRIPTION Suggests.
- Bookdown section numbering disabled per request — section titles
  read as prose, not numbered chapters.
- README quick-start fixed: was referencing example_aoi.gpkg
  (does not exist) and example_catalog.json arg (no longer needed —
  cd_catalog() defaults to the live S3 catalog). Replaced "Vignette
  coming soon" link with the live pkgdown article.

Relates to #39
Relates to NewGraphEnvironment/sred#23

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>