fix: apply consistent surrogate-sanitizing read.

[StarsExpress]

With PR #775 refactoring issue #463's part 2 and just merged into main,
PR #756 fixes #463's part 1: align repeated "read file + sanitize surrogates" block.

---

Repeated code block

with open(file_path, "r", encoding="utf-8", errors="surrogateescape") as f:
    original = f.read()

try:
    original = original.encode("utf-8", errors="surrogatepass").decode("utf-8", errors="replace")

except (UnicodeEncodeError, UnicodeDecodeError):
    pass

Issue #463 cites 4 duplicated call sites (excluding the recently deleted code_puppy/plugins/file_permission_handler.py in 27efe8f):

• 3 in tools/file_modifications.py — _delete_snippet_from_file, _replace_in_file, _write_to_file.
• 1 in tools/file_operations.py — _read_file.

Actual count — 4 + 2 = 6:

• file_modifications.py has 4th site at _delete_file, and 5th site at _delete_file_async.

All 6 repeated calls are covered by this PR.

---

Correction

On issue #463's framing: _read_file doesn't actually have "better fallback" as described — it only uses errors="surrogateescape" on open with zero post-read cleanup, making it currently the least robust of 6 call sites.

3-layer fallback _sanitize_string already existed and was used by 13 other call sites; unifying via read_text_sanitized brings _read_file up to that same standard for the first time.

---

What has changed

• Moves _sanitize_string from code_puppy/tools/file_operations.py to code_puppy/tools/common.py, since file_operations.py already depends on common.py regarding resolve_path import.

• Thus, its TestSanitizeString moves from tests/tools/test_file_operations_coverage.py to tests/tools/test_common_full_coverage.py.

• Defines read_text_sanitized next to atomic_write_text inside code_puppy/tools/common.py.

• Aligns the aforementioned 8 repeated occurrences with read_text_sanitized.

• Adds some blank lines in these 8 repeated sites to enhance code readability.