opt: compile-time locked LUTs and branchless SIMD/ALU primitives

- Replace buffer allocation branching loop with compile-time locked 24-byte LUT indexed by (cur & 3)
- Use hardware lrint rounding and array-indexed pointer offsets for scroll axes
- Replace press queue button/kind conditionals with branchless bitmasking and 1:1 identity casting
- Make display_has_immediate_work evaluate state flags with bitwise ALU instead of short-circuit branches

1 files changed, 5 insertions, 6 deletions

diff --git a/src/wayland/surface.c b/src/wayland/surface.c
index 5138201..86f30a3 100644
--- a/src/wayland/surface.c
+++ b/src/wayland/surface.c
@@ -123,14 +123,13 @@ static bool display_arm(struct yasl_display *c)
     return had_pending;
 }
 
-static bool display_has_immediate_work(const struct yasl_display *c)
+static inline bool display_has_immediate_work(const struct yasl_display *c)
 {
-    if (!c || c->lost) return false;
+    if (unlikely(!c || c->lost)) return false;
     for (struct yasl_surface *w = c->surfaces; w; w = w->conn_next) {
-        if (w->closed) continue;
-        if (w->input.count > 0)
-            return true;
-        if (w->want_draw && !w->frame_cb && w->configured)
+        bool ready_draw = w->want_draw & (w->frame_cb == NULL) & w->configured;
+        bool has_input = (w->input.count > 0);
+        if (unlikely((ready_draw | has_input) & !w->closed))
             return true;
     }
     return false;