fix: pixel-identical rendering (z2d stroker port) and last P3 items

Resolves the 3 P3 items in docs/rewrite/notes/review-rv12-codex.md
(final-Z15.md):
- Sprite strokes: tiny-skia is replaced by a port of z2d 0.10's stroke
  plotter (sprite/stroke.rs) plus z2d's arc segmentation and truncating
  integer compositing, feeding the ported 4x4 multisample rasterizer.
  All 36 reference sprite sheets now match Ghostty's byte for byte (test
  tolerance 0), and `just visual-parity` with a production build shows
  0 differing pixels (max delta 0) against the installed Ghostty 1.3.1
  for the test page (block and bar cursor) and the tabs+splits scenario,
  terminal area and chrome alike. tiny-skia dependency removed.
- zig_error maps errno values per target via libc (Linux, Darwin/BSD)
  and Win32 codes the way Zig std does on Windows.
- theme=. / theme=.. report "it is a directory" via a port of
  std.fs.path.basename, matching Ghostty.

Test Plan:
- cargo test --workspace --no-fail-fast (3498 passed, exit 0)
- cargo build --profile production && just visual-parity
- just check-macos (both), just check-windows, just check-wasm

docs/rewrite/notes/final-Z15.md

@@ -0,0 +1,29 @@
+# Z15: RV12 P3 closures
+
+## 1. Platform error codes (`ghostty-os/src/zig_error.rs`)
+`errno_name` is now per target: unix uses `libc::E*` constants (Linux and Darwin/BSD numbering both
+correct, e.g. Darwin `ELOOP=62`, `ENAMETOOLONG=63`, `EAGAIN=35`); Windows maps Win32 codes the way
+Zig std's `windowsCreateProcess` / file-open paths do (FILE_NOT_FOUND, PATH_NOT_FOUND, DIRECTORY ->
+FileNotFound; ACCESS_DENIED -> AccessDenied; INVALID_NAME, FILENAME_EXCED_RANGE -> NameTooLong;
+SHARING_VIOLATION -> FileBusy; COMMITMENT_LIMIT -> SystemResources; DISK_FULL -> NoSpaceLeft; the
+ENOEXEC family -> InvalidExe), using `windows-sys` constants. Other targets (wasm): `Unexpected`.
+Tests are cfg'd per platform (libc constants everywhere, literal tables for Linux/Apple/Windows) and
+compile-checked with `just check-macos` / `just check-windows`.
+
+## 2. Dot theme names (`ghostty-config/src/theme.rs`)
+`zig_basename` ports `std.fs.path.basename` (ComponentIterator.last: trailing separators stripped,
+`.`/`..` are ordinary names; `X:` drive root on Windows) instead of `Path::file_name`. `theme=.` and
+`theme=..` now pass the basename check and report Zig's `not reading theme from ".": it is a
+directory`. Differentially checked against /usr/bin/ghostty (`.`, `..`, `a/`, `a/b`, `./`: identical).
+
+## 3. Sprite stroking (`ghostty-font/src/sprite/{stroke,canvas}.rs`)
+tiny-skia is gone (dependency removed). New `stroke.rs` ports z2d 0.10 `stroke_plotter`, `Face`, `Pen`,
+`Slope`, `Spline` (curve flattening with round joins), `PointBuffer` and Contour/Polygon edge building
+(identity CTM, no dashes). `canvas.rs` now builds z2d-style nodes (`close` appends the implicit
+`move_to`), fills via direct node flattening in f64 (no f32 skia round trip), ports z2d's Cairo-style
+`arc` (segment count from the tolerance table), and uses z2d's truncating integer `src_over`
+(`s + d - s*d/255`, `dst_in` `cov*a/255`) instead of rounding.
+Sprite-sheet reference test: every one of the 36 sheets is now bit-exact (was 1.2% differing, max 86);
+the test now requires 0 differing pixels.
+`just visual-parity` (iteration build of the production profile): page, page-bar, tabs and chrome all
+0 differing pixels (was 196 px / 0.020% for page).

docs/rewrite/notes/visual-parity-VP.md

@@ -30,13 +30,13 @@
 
 | scenario | before | after |
 |---|---|---|
-| page (full frame) | 0.130% (1250 px, max 93) | 0.020% (196 px, max 86; 0.001% > 32 levels) |
-| page-bar | same as page | 0.020% |
+| page (full frame) | 0.130% (1250 px, max 93) | 0.000% (0 px; was 0.020% / 196 px before Z15) |
+| page-bar | same as page | 0.000% |
 | tabs + splits | 44.6% (a whole pane blank, bug) | 0.000% (bit-identical) |
 | chrome (headerbar, tab bar) | 0.000% | 0.000% |
 
-Sprite sheets (`ghostty-font` tests, Zig reference PNGs): differing pixels 5.76% -> 1.20%, pixels off by
-more than 96 levels 74 -> 0 (test tolerance tightened from 4% to 0.1% big diffs).
+Sprite sheets (`ghostty-font` tests, Zig reference PNGs): differing pixels 5.76% -> 1.20% -> 0 (bit-exact after Z15;
+test tolerance is now zero).
 
 Font metrics, cell size, baseline, glyph rasterization, colors, cursor, underline positions, padding,
 GTK CSS/headerbar/tab bar were already pixel-identical; no change was needed there.
@@ -51,12 +51,12 @@ GTK CSS/headerbar/tab bar were already pixel-identical; no change was needed the
 2. **Sprite antialiasing** (`ghostty-font/src/sprite/canvas.rs`): replaced tiny-skia's coverage with a port of
    z2d 0.10's default `multisample_4x` rasterizer (4x4 samples, scanline spans with edge x rounded at
    sample row centers, non-zero filter, alpha = `coverage * 16 - 1`, full = 255, z2d's Bezier
-   flattening with tolerance 0.1). Stroke outlines still come from tiny-skia's stroker.
+   flattening with tolerance 0.1). (Stroke outlines were still tiny-skia's until Z15, which ported z2d's stroker.)
 
-## Remaining differences (page: 196 px, all <= 86 levels)
+## Remaining differences
 
-- Curly underline: first/last pixels and peaks differ slightly. z2d builds its own stroke polygon (pen/joins)
-  for curves; tiny-skia's stroker offsets the Bezier differently.
-- Diagonal box glyphs (`U+2571..2573`) have a few edge pixels off by up to ~38 levels (same cause).
-- Porting z2d's stroker (`tess/stroke_plotter`, `Pen`) would remove the rest; judged not worth it
-  (0.02% of pixels, imperceptible).
+None (Z15). z2d's stroke plotter (pen, joins, caps, spline flattening), Cairo-style arc segmentation and
+z2d's truncating integer compositing were ported (`ghostty-font/src/sprite/stroke.rs`, `canvas.rs`;
+see `final-Z15.md`). Final `just visual-parity`: page, page-bar and tabs are 0 differing pixels on the
+full frame, the terminal area and the chrome (max delta 0). All sprite sheets in the `ghostty-font`
+reference test are bit-exact (tolerance now zero).