improved slice.sort_from_permutation_indices

[TheRadischen]

Old Implementation

follows the chain of indices each iteration until a new index is hit, effectivly doing the same thing multiple times.
performs 2x more data moves than neccessary

New Implementation

immediatly moves the data into place following the indices chain, by using a temp variable
then skips indices it has allready visited by swapping their sign
the indices array is unchanged at the end

Benchmark

https://github.com/TheRadischen/odin-utils/blob/main/sort/test/new_permutaiton/test.odin on new branch
2 - 8x speed improvement for not much more complicated logic

Size: 10
new perm on 80 byte 100ns
old perm on 80 byte 200ns
new perm on 8 byte 100ns
old perm on 8 byte 200ns

Size: 100
new perm on 80 byte 400ns
old perm on 80 byte 1.3µs
new perm on 8 byte 400ns
old perm on 8 byte 1.4µs

Size: 1000
new perm on 80 byte 2.1µs
old perm on 80 byte 13.1µs
new perm on 8 byte 2.8µs
old perm on 8 byte 14µs

Size: 10000
new perm on 80 byte 56.3µs
old perm on 80 byte 242µs
new perm on 8 byte 110.3µs
old perm on 8 byte 338.8µs

Size: 100000
new perm on 80 byte 1.5164ms
old perm on 80 byte 6.6669ms
new perm on 8 byte 2.0299ms
old perm on 8 byte 7.1748ms

Size: 1000000
new perm on 80 byte 89.2422ms
old perm on 80 byte 739.469ms
new perm on 8 byte 86.1281ms
old perm on 8 byte 781.5958ms