Ensure small-sort is inlined

This greatly improves binary-size. With the recent change we made sure
to only call the small-sort once. While we don't expect the
stable_quicksort loop to be inline-able because it is recursive. We want
to be on the safe side and annotate it.

src/quicksort.rs

@@ -10,6 +10,7 @@ use crate::smallsort::SmallSortTypeImpl;
 ///
 /// `limit` when initialized with `c*log(v.len())` for some c ensures we do not
 /// overflow the stack or go quadratic.
+#[inline(never)]
 pub fn stable_quicksort<T, F: FnMut(&T, &T) -> bool>(
     mut v: &mut [T],
     scratch: &mut [MaybeUninit<T>],

src/smallsort.rs

@@ -25,6 +25,7 @@ pub trait SmallSortTypeImpl: Sized {
 impl<T> SmallSortTypeImpl for T {
     default const SMALL_SORT_THRESHOLD: usize = 16;
 
+    #[inline(always)]
     default fn sort_small<F: FnMut(&T, &T) -> bool>(
         v: &mut [T],
         _scratch: &mut [MaybeUninit<T>],
@@ -44,6 +45,7 @@ where
 {
     const SMALL_SORT_THRESHOLD: usize = 20;
 
+    #[inline(always)]
     fn sort_small<F: FnMut(&T, &T) -> bool>(
         v: &mut [T],
         scratch: &mut [MaybeUninit<T>],