fix(zim): correct config_indexing docstring to full-text only

The docstring of Creator.config_indexing claimed it toggled both
full-text and title indexing, but libzim only toggles the full-text
index; the title index is always built. Update the docstring to
reflect libzim's actual behaviour and add a regression test.

Author: bugfix-mission

CHANGELOG.md

@@ -21,6 +21,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Fix outdated system dependencies in README: remove Pillow build-time deps (bundled in wheels), add missing `libcairo` across all platforms (#152)
 - Improve contribution setup instructions: use `hatch shell`, clarify commands must be run from local clone root (#153)
 - Bring coverage back to 100% (#293)
+- Fix `Creator.config_indexing` docstring to reflect that only the full-text index is toggled; title indexing is always performed by libzim (#294)
 
 ### Changed
 

src/zimscraperlib/zim/creator.py

@@ -127,9 +127,12 @@ def __init__(
     def config_indexing(
         self, indexing: bool, language: str | None = None  # noqa: FBT001
     ):
-        """Toggle full-text and title indexing of entries
+        """Toggle full-text indexing of entries
 
-        Uses Language metadata's value (or "") if not set"""
+        Uses Language metadata's value (or "") if not set.
+
+        Note: title indexing is always performed by libzim and cannot be
+        disabled via this method; only the full-text index is toggled."""
         language = language or self._get_first_language_metadata_value() or ""
         if indexing and not is_valid_iso_639_3(language):
             raise ValueError("Not a valid ISO-639-3 language code")