build: D1 cross-compile via cargo-zigbuild (4 non-macOS targets)

Replace the single-target musl cc with cargo-zigbuild + zig in the flake
devShell — one universal cross cc/linker (incl. rusqlite's bundled SQLite
C) for all four non-macOS D1 targets, added to rust-toolchain.toml:
  x86_64/aarch64-unknown-linux-musl              (static, D2)
  x86_64-pc-windows-gnu, aarch64-pc-windows-gnullvm  (standalone .exe)

Windows links -lsynchronization (std WaitOnAddress), which rust-overlay's
toolchain and zig's mingw don't ship; the symbols are forwarded by
kernel32, so an empty stub libsynchronization.a (ci/winstub/, wired via
.cargo/config.toml for the windows targets only) satisfies the linker.
Verified: all four build; linux static; windows valid PE32+.

ci/winstub/README.md

@@ -0,0 +1,30 @@
+# `ci/winstub/` — empty Windows import-lib stub
+
+`libsynchronization.a` here is an **empty `ar` archive** (8 bytes: `!<arch>\n`),
+referenced by `.cargo/config.toml` via `-L native=ci/winstub` for the Windows
+release targets.
+
+## Why
+
+The D1 release matrix cross-compiles Windows binaries from Linux with
+`cargo zigbuild` (see `docs/ci/adr/`). Rust's `std` links `-lsynchronization`
+for its `WaitOnAddress`-based thread parking. That import library is normally
+provided by Rust's `rust-mingw` "self-contained" component — which `rust-overlay`
+does not ship — and Zig's bundled mingw doesn't carry it either, so the link
+fails with:
+
+```
+error: unable to find dynamic system library 'synchronization'
+```
+
+The functions it would import (`WaitOnAddress`, `WakeByAddressSingle`,
+`WakeByAddressAll`) are **forwarded by `kernel32.dll`**, which is already linked,
+so they resolve at link and run time without a real `synchronization` import
+library. An **empty** stub is therefore sufficient: it satisfies the `-l`
+lookup and contributes no symbols.
+
+## Regenerating
+
+```
+zig ar rcs ci/winstub/libsynchronization.a
+```