Optimize Quick Sort, N-Sum, and fix bug in Is Palindrome

[DhruvrajSinhZala24]

1. Quick Sort: Added Median-of-Three pivot selection and Insertion Sort for small subarrays.\n2. N-Sum: Refactored to reduce redundant sorts and copies using recursion with indices.\n3. Is Palindrome: Fixed potential IndexError and improved efficiency.

[keon]

Verified locally: test_array.py, test_sorting.py, and test_string.py all pass (116/116). Algorithmically the changes look fine — median-of-three + insertion-sort threshold for quicksort is a standard, well-motivated optimization, and the n-sum refactor to indices rather than slicing is a real improvement.

A few process notes for the maintainer:

1. Bundling. Three unrelated changes (quicksort optimization, n-sum refactor, is_palindrome bugfix) in a single PR makes review and revert harder. Splitting into three PRs would be much friendlier to review.

2. Overlap with #2749. The is_palindrome portion duplicates #2749, which is a tighter, more focused fix to the same root cause (_remove_punctuation stripping digits). I'd suggest landing #2749 first and dropping the is_palindrome changes from this PR.

3. n-sum behavior. The refactor drops the sorted([...]) call on each appended tuple that the original _append_elem_to_each_list had. The new code relies on the input being sorted (which it is, via the upstream nums.sort()) so output order is preserved — that's correct, but worth a one-line comment in the code to flag the invariant for future readers.

Not blocking — just suggesting a cleaner split.