Pylangacq

Fixed
- Handled the duration mark (e.g., `[ 0.4]`) in utterance cleaning.

Added
- Added support for Python 3.12.
- Handled pre-clitics and post-clitics from %mor tiers
and honored their distinction in the parsed utterance.

Added
- Added support for Python 3.11.

Changed
- Updated the test data from Brown's Eve from the upstream CHILDES.

Removed
- Dropped support for Python 3.7.

Added
- Added the `exclude_switch` option for MLU (`mlu()`, `mlum()`, and `mluw()`),
so that words with `s` for switching language may be excluded.

Fixed
* Fixed MLU computation (`mlu()`, `mlum()`, and `mluw()`):
- If xxx, yyy, or www appears in an utterance, the whole utterance is ignored.
- If there are no MLU-relevant words/morphemes in an utterances, the whole utterance is ignored.

Fixed
* Moved `download_and_extract_brown` test function to under the `pylangacq` package
namespace, as tests from `BaseTestCHATReader` require downloaded CHAT data files.

Changed
* Restructured the repository to use top-level `src/` and `tests/` directories.

Removed
* Removed `setup.py`.

Fixed
* Moved `BaseTestCHATReader` back under the `pylangacq` package namespace
so that downstream packages can import `BaseTestCHATReader` for testing.