Add median

[vil02]

This PR adds a function computing the median (in the naive/basic way).

[appgurueu]

It should be possible to do this in (expected) linear time - without a much more complex implementation - by reusing the existing quickselect implementation, specifically the one using median of medians if you want guaranteed linear time (see here for an example usage of quickselect using median of medians).

To use this, you'd probably want to first shallowcopy the array (since quickselect mutates it).

Then the case of an odd number of elements would be easy enough: Just local i = quickselect(copy, math.ceil(#copy / 2)); return copy[i].
For an even number of elements, you'd still do local i = quickselect(copy, math.ceil(#copy / 2)). Since this partitions copy, you can now do local j = quickselect(copy, 1, nil, i + 1), to get the next largest element after the middle element to the right. (You could of course also just save the element at copy[i], then do a local j = quickselect(copy, #copy/2 + 1)).

[vil02]

My original plan was to add median.lua and quick_median.lua. I overlooked the fact that the second one it is basically implemented. I will update this PR.

[appgurueu]

Changed {unpack(list)} to a loop because unpack(list) puts the items on the stack, which is problematic for large lists (the stack is limited and can overflow; on 5.1 / JIT it's pretty "small", about 16k / 65k). Also removed the else since it is redundant with the early return.