feat(pointer): add an outline tone with a transparent fill and white edge

[LeadcodeDev]

Refs #344 — its point 5.

PointerTone was light | dark, both filled. A reference video uses a pointer that is a white outline over a transparent fill. PointerTone::Outline adds it.

Why a third tone rather than a fill field

tone exists precisely so a scene picks one coherent word instead of two matched hex values, and colors() already dispatches fill and outline from a single match arm — a third arm extends that uniformly. A separate fill field would need its own "no fill" semantics, interact awkwardly with the existing color/outline_color overrides, and reintroduce exactly the two-hex-values problem tone was built to avoid. The override mechanism keeps working: a scenario can still set color explicitly for a translucent fill under an outline pointer.

A bug found on the way

The click ring defaulted to the arrow's fill colour. Under Outline, fill is "transparent" — which parses to rgb (0,0,0) with alpha supplied separately at draw time — so the ring would have rendered near-black under a white outline. Not a hard-coded bug, but visually incoherent, and it would have shipped looking like the feature was broken.

ring_fallback_color(&self, fill, outline) threads the already-resolved strings: Outline falls back to outline (white, or the author's outline_color override), Light and Dark keep falling back to fill exactly as before.

Verification

Four tests, all pixel-level:

• deep_interior_point_is_well_clear_of_the_outline_stroke — the probe point is found via Path::contains plus a margin check, not hand-picked coordinates, so the real test cannot pass by sampling the wrong place
• outline_tone_leaves_the_interior_transparent_and_paints_a_pointer_coloured_edge — interior must equal the background exactly (0,128,0,255), edge must be solid white
• filled_tones_still_paint_pointer_colour_in_both_interior_and_edge — pins Light and Dark byte-for-byte, the backward-compatibility guard
• outline_tone_keeps_the_click_ring_visible_and_matched_to_the_white_outline

Both proven to catch the absence:

# Outline painted filled instead of transparent
an outline pointer must let the background show through its interior,
got (255, 255, 255, 255)
  left: (255, 255, 255, 255)
 right: (0, 128, 0, 255)

# ring_fallback_color forced back to always returning fill
the click ring on an outline pointer must default to the outline's white,
not a hard-coded or transparent-derived colour, got (9, 9, 9, 255)

cargo fmt --all --check, cargo clippy --workspace --all-targets -- -D warnings, cargo test --workspace (1529) all clean.

SKILL.md needs no change: it references pointer only by component name and links to pointer-walkthrough.md, which this PR extends. It never enumerated the tone values.

Written comment-free, per the codebase-wide rule from #345.