Built-in sorting algorithms from standard libraries are considered to
cost O(1) auxiliary space as heapsort[1] exists, even though major
implementations, such as Java's Arrays.sort[2,3]'s dual-pivot
quicksort[4] and Python's list.sort[5,6]'s powersort[7] actually need
O(log n).

[1] <https://dl.acm.org/doi/pdf/10.1145/512274.3734138>
[2] <https://docs.oracle.com/en/java/javase/24/docs/api/java.base/java/util/Arrays.html#sort(int%5B%5D)>
[3] <https://github.com/openjdk/jdk/blob/c62223a5af747bc5cbdd3d970dd994f74aa08834/src/java.base/share/classes/java/util/DualPivotQuicksort.java#L242-L394>
[4] <https://www.kriche.com.ar/root/programming/spaceTimeComplexity/DualPivotQuicksort.pdf>
[5] <https://docs.python.org/3/library/stdtypes.html#list.sort>
[6] <https://github.com/python/cpython/blob/8e8786f8986353e20c1c4406c34409a6139fa073/Objects/listobject.c#L2874-L3163>
[7] <https://www.wild-inter.net/publications/munro-wild-2018.pdf>