fix: _remove_punctuation strips digits causing false palindrome results

[faizzyhon]

Problem

_remove_punctuation() in algorithms/string/is_palindrome.py used ascii_letters (a-z, A-Z only) to filter characters, silently discarding all digit characters (0-9).

Four of the five palindrome-check variants — is_palindrome_reverse, is_palindrome_two_pointer, is_palindrome_stack, and is_palindrome_deque — all call this helper, so they all returned wrong results for strings containing digits:

# Before this fix:
is_palindrome_reverse("a1b2a")    # True  ← WRONG  (digits stripped → "aba" = palindrome)
is_palindrome_reverse("1a2")      # True  ← WRONG  (digits stripped → "a"   = palindrome)
is_palindrome("a1b2a")            # False ← correct (uses .isalnum() directly)

Fix

Replace the ascii_letters membership test in _remove_punctuation with str.isalnum(), which correctly retains digit characters:

- return "".join(char.lower() for char in text if char in ascii_letters)
+ return "".join(char.lower() for char in text if char.isalnum())

Remove the now-unused from string import ascii_letters import. Update all five functions' docstrings to say "alphanumeric" (was "alphabetic"). Add bound-guards left < right to the two inner while-loops in is_palindrome to prevent an infinite loop on all-punctuation input.

Tests

Added test_is_palindrome_with_digits which exercises all five variants with:

• "12321" → True (valid digit palindrome)
• "a1b2a" → False (not a palindrome once digits are kept)

All 567 existing tests continue to pass.

[keon]

APPROVED. Verified locally: test_string.py 68/68 pass. Good catch — replacing the ascii_letters membership test with str.isalnum() matches the behavior of the standalone is_palindrome function (which already used .isalnum()), so all five variants now agree. The left < right bound-guards on the inner while-loops in is_palindrome are a nice incidental hardening against all-punctuation input.

Note for the maintainer: this overlaps with the is_palindrome portion of #2745.