fix: Boyer-Moore bad character shift was dead code in for-loop

The bad_character_heuristic() method used a for-loop with an
assignment to the loop variable i, which was immediately
overwritten by the next iteration. This caused the algorithm
to degrade from O(n/m) to O(n*m) naive search.

Changed to a while-loop so the shift actually takes effect.
Added max(i+1, shift) guard to prevent backward skips when
the mismatched character appears to the right of the mismatch
in the pattern. Added edge case doctests.

Author: anyncfunction

strings/boyer_moore_search.py

@@ -83,18 +83,30 @@ def bad_character_heuristic(self) -> list[int]:
         >>> bms = BoyerMooreSearch(text="ABAABA", pattern="AB")
         >>> bms.bad_character_heuristic()
         [0, 3]
+
+        >>> bms = BoyerMooreSearch(text="AAAAA", pattern="AB")
+        >>> bms.bad_character_heuristic()
+        []
+
+        >>> bms = BoyerMooreSearch(text="ABABAB", pattern="ABA")
+        >>> bms.bad_character_heuristic()
+        [0, 2]
+
+        >>> bms = BoyerMooreSearch(text="", pattern="AB")
+        >>> bms.bad_character_heuristic()
+        []
         """
 
         positions = []
-        for i in range(self.textLen - self.patLen + 1):
+        i = 0
+        while i <= self.textLen - self.patLen:
             mismatch_index = self.mismatch_in_text(i)
             if mismatch_index == -1:
                 positions.append(i)
+                i += 1
             else:
                 match_index = self.match_in_pattern(self.text[mismatch_index])
-                i = (
-                    mismatch_index - match_index
-                )  # shifting index lgtm [py/multiple-definition]
+                i = max(i + 1, mismatch_index - match_index)
         return positions