pub const PARALLEL_MIN_WORK: usize = _; // 524_288usizeExpand description
Parallel regions with less work than this, in f64 operations, run on the
calling thread (dispatching to the pool would cost more than it saves).
On an M1 Pro, compute_sve was fastest at about this threshold both in
f64 (ε ≥ 10⁻⁸) and, counting a double-double operation as 16 f64
ones, in double-double.