fix: reduce palindrome_partitioning time complexity from O(n^3) to O(n^2)

[https-hari]

Description of Change

The existing implementation claims O(n^2) time complexity in its documentation
but actually runs in O(n^3) due to three nested loops:

• Outer loop over lengths: O(n)
• Inner loop over start indices: O(n)
• Innermost loop over partition points: O(n)

Fix

Replaced the 2D cuts table with a 1D DP array cuts[i], where
cuts[i] = minimum cuts needed for str[0..i].

For each position i, we scan j from 1 to i — if str[j..i] is a palindrome,
then cuts[i] = min(cuts[i], cuts[j-1] + 1). This eliminates the third
nested loop entirely, genuinely achieving the O(n^2) complexity the
original code claimed but never delivered.

Complexity

The original had three nested loops making it O(n^3) in time despite claiming
O(n^2) in the docs. This fix brings it down to true O(n^2) by eliminating
the innermost partition loop entirely. Space stays O(n^2) for the palindrome
table, plus a small O(n) for the new 1D cuts array.

Tests Added

• Original 3 test cases preserved and passing
• Added 7 new edge cases: empty string, single char, two same chars,
  two different chars, full palindrome, all same chars, no palindrome substrings

Relates to #2456