mitchell's touchups

- benchmark: avoid buffers to avoid a memcpy
- build: keep frame pointers on macOS. There was some debug changes from
  Zig 0.15 and this helps. Also, Apple actually requires/expects x29 to
  always be a frame pointer.
- build/macos: force libSystem symbols instead of compiler-rt
- global: add InitOpts.tool so that ghostty-gen/bench can parse their
  own actions in `+action`
- quirks: provide our own vectorized memset. see the comment for more
  details why.
- synthetic: fix UB by accessing global.io before it was initialized
- terminal/hash_map: force inline for unique repr types. Zig 0.15
  inlined and 0.16 doesn't, measured a huge slowdown in hyperlink
  benchmarks.
- terminal: add explicit `@Vector` usage for storing a run of identical cells
  as well as for scanning printable cells. This auto-vectorized in Zig
  0.15 but not in Zig 0.16. This produces the same assembly.
- unicode: properties and LUT need power-of-two backing integer to avoid
  bad LLVM codegen

src/simd/main.zig

@@ -9,6 +9,18 @@ pub const index_of = @import("index_of.zig");
 pub const vt = @import("vt.zig");
 pub const codepointWidth = codepoint_width.codepointWidth;
 
+/// The number of vector lanes to use for manually vectorized hot
+/// loops operating on elements of type T, or null if the target has
+/// no usable SIMD support and a plain scalar loop should be used
+/// instead (e.g. wasm32 without simd128).
+///
+/// This is twice the native vector width so that each iteration works
+/// two vector registers, giving better instruction-level parallelism.
+pub fn lanes(comptime T: type) ?comptime_int {
+    const native = std.simd.suggestVectorLength(T) orelse return null;
+    return native * 2;
+}
+
 test {
     @import("std").testing.refAllDecls(@This());
 }