//! The one pid → lane mapping, shared by every component that keeps one
//! process's work ordered while independent processes run concurrently: the
//! scheduler task lanes, the per-process event gate and the store writer
//! shards. Sharing it means the routing can never disagree — and a component
//! that must keep its work serialized against another's (scheduler jobs
//! against the event handlers of the same process) gets it for free.
/// Index of the lane `pid` belongs to, in `0..lanes` (`lanes` is never 0; a
/// count below one is treated as one). FNV-1a over the pid, so the same pid
/// always maps to the same lane within a process.
pub