load gaze files with gaze.from_csv()

[dkrako]

Description of the problem

Currently we need the Dataset class to load GazeDataFrames.

Description of a solution

It would be much nicer to have a self-contained solution in a function like pymovements.gaze.from_csv() which returns a GazeDataFrame loaded from a csv file.

Most functionality is already included in dataset/dataset_files.py, so parts can be reused in gaze/io.py.
We won't need to add any fileinfo columns for now.
We will refactor dataset/dataset_files.py after finishing #508 and #518.

For a start, we can use **kwargs to forward all specific reading arguments to polars.read_csv()

Simply passing **kwargs will also shorten testing and documentation. But we should discuss if we want to add all the arguments explictly.
Although having all arguments explicit is probably the best solution, we would then need to keep the documentation up to date. We should therefore discuss if a more self-contained documentation is worth the effort. (probably it is)

Minimum acceptance criteria

• pm.gaze.from_csv() in gaze/io.py
• tests in tests/gaze/io/csv_test.py
• create at least one monocular and a binocular csv test file with 10 rows (can be all zero) in tests/gaze/io/files/

Code example

proposed function signature:

def from_csv(
        file: str | Path,
        experiment: Experiment | None = None,
        *,
        trial_columns: list[str] | None = None,
        time_column: str | None = None,
        pixel_columns: list[str] | None = None,
        position_columns: list[str] | None = None,
        velocity_columns: list[str] | None = None,
        acceleration_columns: list[str] | None = None,
        **read_csv_kwargs,
) -> GazeDataFrame:

[dkrako]

the PR should include simple csv test files which should be then loaded in tests (maybe also write a fixture to simplify usage).

there should actually be a range of csv files, although the files should probably all be limited to about 10 rows
we can also test several header configurations (monocular, binocular, with/without time column, pixel data, dva data, stimulus, etc.)

this will probably help resolving #492 by a lot, because we could also create toy version csv files for gazebase and judo1000