Extend data loading functions to include extra trials from the WAR and/or Move task

[ChesterHuynh]

Extend functions in io/read.py to allow inclusion of fixation data from the WAR task

[adam2392]

I am extending the data structure for "Trials" to contain more of the metadata information. In addition, I've written the behavioral XY coordinate data over time as a compressed "tsv" file. Will upload an API to read this data.

[adam2392]

The strategy to robustly get metadata to the Epochs data structure and downstream tasks is:

1. Load in Epochs (e.g. 98 x T): T time points
2. Load in trial metadata (e.g. 98 x M): M metadata points
3. Set metadata for Epochs data structure: See https://mne.tools/stable/generated/mne.Epochs.html for things you can thus do at the epochs level.

Dropping epochs -> results in dropping said metadata All metadata goes hand-in-hand w/ the epochs

Resulting in the following API:

• read_move_trial_epochs
• read_trial_metadata
• read_behav_xy_coords

and the dataclass:

• Trial specifying metadata for a specific trial, with a to_data_frame method.

Note on XY coordinate data

The XY coordinate movement data of the cursor has the same sampling rate as the raw EEG. Thus we can treat it like a mne.io.Raw object with two sensors: (x, and y). We can add the corresponding events data structure to it, so we could theoretically form an Epochs data structure. Or we can perform parsing on it to determine average speed between two endpoints of a specific trial.

• if we only interested in the speed in the same interval as the Epochs, we can epoch the XY coordinate data (easy; same procedure of epoching as Raw EEG data)
• if we want some general data, such as average speed between two event markers (e.g. Go Cue and Hit Target), then we must compute this without using Epochs since time duration between these two events for each trial differs. We'll do something like a list comprehension "[trial.compute_speed() for trial in xy_raw.trials]"