adding logger args

[ysjprojects]

Addresses #1909.

More fine-grained logging args: supports e.g wandb args specifying run_name and group_name for now.

New LoggerArgs as it fits neither TrainArgs nor EvalArgs. We can expand on this in the future.

[Borda]

wandb is not the only one or default logger so we shall not name it with generic LogArgs

[ysjprojects]

Removed references to wandb in LogArgs.

Also, is there a reason why we don't pass the name argument to csv and tensorboard loggers?

def choose_logger(
    logger_name: Literal["csv", "tensorboard", "wandb", "mlflow"],
    out_dir: Path,
    name: str,
    log_interval: int = 1,
    log_args: Optional[Dict] = None,
    resume: Optional[bool] = None,
    **kwargs: Any,
):
    if logger_name == "csv":
        return CSVLogger(root_dir=(out_dir / "logs"), name="csv", flush_logs_every_n_steps=log_interval, **kwargs)
    if logger_name == "tensorboard":
        return TensorBoardLogger(root_dir=(out_dir / "logs"), name="tensorboard", **kwargs)

@Borda