Vignette: add tmax/tmin with ecological framing + clip maps to watershed group

[NewGraphEnvironment]

Problem

The cd package now ships tmax/tmin on STAC (unified EDH-derived grid, 1950-2025) but the current vignette only narrates tmean and soil_moisture. tmax/tmin carry distinctive climate signals that tmean hides:

• Overnight lows (tmin) warm faster than daytime highs (tmax) under climate change — one of the clearest regional fingerprints
• Diurnal range (tmax - tmin) is a separate trend affecting evapotranspiration, fire weather, and ecosystem stress
• Specific thresholds tied to biology use tmax/tmin directly (frost days, growing-degree days, heat-stress days)

Proposed content

One or two additional sections/chunks in vignettes/climate-departure.Rmd:

• tmin warming figure: annual tmin anomaly for the AOI, compared side-by-side with tmean. Commentary on "nights are warming faster than days" as a clear departure signal.
• Diurnal range trend: compute tmax - tmin per year, plot trend. Brief commentary.
• Optional: threshold-based derived metric — frost days (tmin < 0°C), heat days (tmax > 30°C), or growing-degree days for a biologically relevant AOI.

Keep it short — the vignette is long enough. Goal is to showcase that the package carries and supports tmax/tmin as first-class citizens, not to duplicate every analysis twice.

Out of scope

• Adding new cd functions (existing cd_extract / cd_anomaly / cd_compare already handle tmax/tmin once the data is in the catalog, which it now is).
• Deeper narrative restructuring.

Tracking

Depends on #36 (EDH migration) landing so the tmax/tmin COGs are on S3 before the vignette references them. Once merged, this issue can be picked up any time.

Relates to #36
Relates to NewGraphEnvironment/sred-2025-2026#23

[NewGraphEnvironment]

Scope expansion 2026-04-14: map clipping + deeper ecological framing

Broadening this issue to cover two related vignette improvements that are logical to do together.

1. tmax/tmin narrative with ecological why

The original scope covered adding tmax/tmin figures. Expanding to make sure the prose tells why these specific variables matter ecologically — not just "temperature is warming" but the separable signals in daytime vs overnight change.

Specific angles to weave in (pick 2-3, don't cram all):

• Overnight lows (tmin) warming faster than daytime highs (tmax) — one of the clearest ERA5-observed regional fingerprints of climate change. Shrinking diurnal range (tmax - tmin) has direct consequences for evapotranspiration, cold-air pooling in valleys, and overnight cooling during summer heat events.
• Frost days (tmin < 0 °C) and growing-degree-days — biologically meaningful thresholds for phenology, pest range expansion (e.g., mountain pine beetle overwintering survival), and agriculture.
• Heat stress days (tmax > ecologically-relevant threshold) — fish thermal stress in streams lags air temperature, but tmax trends set the envelope. For BC salmon watersheds this is the temperature question that actually matters for management.
• Vapour pressure deficit (vpd) × tmax — compound drought-stress signal during hot-dry summers. Tmax warming is often coincident with humidity declines, amplifying plant water stress.

Keep it tight — 1 figure per angle, 1-2 paragraphs of commentary each. Goal is to showcase tmax/tmin as first-class citizens in the package and land the "departure isn't just about means" point.

2. Map clipping to watershed group

The existing maps in the vignette (AOI overview, spatial anomaly raster overlays, etc.) show BC-wide or coarsely-clipped streams, lakes, and context layers. This reads loose and dilutes the story.

Proposed cleanup:

• Clip streams, lakes, roads, BEC zones (whatever context layers the vignette uses) to the watershed group of the example AOI, not the bounding box or province
• For mapgl interactive maps: add the watershed group polygon as a faint boundary, use maxzoom / fitBounds to the AOI, don't render features outside the watershed group
• For tmap static figures: clip all vector layers via sf::st_intersection(x, wsg) before plotting
• Consider adding the watershed group as a styled outline so readers understand the spatial scope at a glance

This lines up with the cartography conventions (gq styles, legend placement, AOI-focused framing).

Shared deliverable

One PR that updates vignettes/climate-departure.Rmd with:

• Cleaner clipped maps for every existing figure
• New tmax/tmin figure(s) with ecological framing
• Keep the vignette rendering time sane (≤ 2 min on CI)

Dependencies

• Watershed group for the AOI needs to be pulled from fwapg (or wherever the current AOI is sourced). Worth a helper if the pattern isn't there already.

Out of scope

• New cd package functions. Existing cd_extract / cd_anomaly / cd_compare already handle tmax/tmin.
• Changing the example AOI — keep Kootenay Lake watershed.

Relates to #36
Relates to NewGraphEnvironment/sred-2025-2026#23