Incorporate the result of `everyvoice check-data` to be able to filter filelists

### Description

#### Add Outlier Filtering to use in combination with `everyvoice check-data`

Currently, the everyvoice check-data command outputs a spreadsheet with all of the data features. It would be helpful to add functionality that allows users to define thresholds to automatically filter outliers from the filelist based on numbers calculated by `everyvoice check-data`.

#### Proposed Feature

Add support for threshold-based filtering, such as:

- Generating a filtered file list based on defined thresholds.

- Example use case:

- Exclude files where speaking_rate > X

- Exclude files where total_loss > Y

This would streamline the process of cleaning data before training and make the output of check-data more actionable.