//! GPU light-culling compute pass.
//!
//! Builds a per-froxel light list (3D grid of `(tile_x, tile_y, z_slice)`
//! cells, exponential view-space depth slices) consumed by the
//! transparent shader and — when the per-mesh CPU bucket marks a mesh
//! as oversized — by the opaque shader's oversized-mesh override path.
//!
//! The cull runs in two stages (`cs_tile` → `cs_main`).
pub use ;