perf: fix deferred divide precision, threadlocal TLS, MSVC target

Removed deferred divide from both ray-tri functions -- the det-scaled
epsilon comparisons lose precision for near-parallel rays, potentially
accepting triangles the original rejects. Downstream SetupBoxFrustum
then writes to wrong globals near 0xCE6738 (SGroupPtr), causing
ERROR #124 on exit.

inflate_hook: replaced manual FS:0x24 thread ID pool with Zig native
threadlocal. Fixes potential crash on Wine where FS segment layout
may differ.

Build: switched from GNU to MSVC ABI (4-5K smaller per DLL, no
.eh_frame unwind tables). Added setjmp.h stub for libdeflate MSVC
build. Added noperf build variant (zig build noperf).

src/bench/main.zig

@@ -241,9 +241,10 @@ pub fn main() void {
     // AddToSpatialGrid -- linked list requires game state, A/B test in-game only
     // bench_addToSpatialGrid();
 
-    if (false) { // disabled: not testing these right now
     bench_collisionDetection();
     bench_rayTriIndexedInt();
+
+    if (false) { // disabled
     bench_entityUpdate();
     }