Fix CHANGELOG cross-refs

Author: znicholls

changelog/19.breaking.md

@@ -1,3 +1,3 @@
 - Required `db_dir` to be passed when initialising [pandas_openscm.db.reader.OpenSCMDBReader][]. This is required to support portable databases
-- Renamed `out_column_type` to `out_columns_type` in [pandas_openscm.db.io.load_timeseries_csv][] for consistency with the rest of the API
+- Renamed `out_column_type` to `out_columns_type` in [pandas_openscm.io.load_timeseries_csv][] for consistency with the rest of the API
 - Bumped the minimum supported version of [filelock](https://py-filelock.readthedocs.io/) to 3.12.3, as only this version handles automatic creation of directories for the lock

changelog/19.feature.md

@@ -1,3 +1,3 @@
 - Made the database portable by only storing relative paths in the file map. This allows the database to be converted to an archive with [pandas_openscm.db.OpenSCMDB.to_gzipped_tar_archive][] and then unpacked elsewhere with [pandas_openscm.db.OpenSCMDB.from_gzipped_tar_archive][]
 - Added [pandas_openscm.db.path_handling][] to clarify how we handle paths internally to support portability
-- Added support for specifying the name of the output columns via [pandas_openscm.db.OpenSCMDB.load][], [pandas_openscm.db.reader.OpenSCMDBReader.load][] and [pandas_openscm.db.io.load_timeseries_csv][]
+- Added support for specifying the name of the output columns via [pandas_openscm.db.OpenSCMDB.load][], [pandas_openscm.db.reader.OpenSCMDBReader.load][] and [pandas_openscm.io.load_timeseries_csv][]