[electrophysiology_browser] Selectable channel types

[MaximeBICMTL]

Description

This PR adds the ability to filter channels displayed in the electrophysiology browser by channel types.

Architecture

• Previously, the electrophysiology browser browser only has access to the channels specified in the chunked index.json files, who contain some information about the channel extracted from the raw files with the mne Python library.
• The channel types for EEG/MEG can be described the channels.tsv BIDS file, which is read by LORIS and saved in the physiological_channel table.
• This PR adds a LORIS electrophysiology API call in the electrophysiology browser to get the channel information from these BIDS metadata files.
• Based on this information, the displayed channel types can be selected in a new "Channel types" dropdown.
• It is possible that an acquisition does not have a channels.tsv file, or that the BIDS channel names do not match the raw channel names. In that case, the relevant channels are just grouped under the "Unknown" channel type, making this feature non-breaking for existing datasets.
• Since this feature affects which channels are displayed on screen, it interacts with the pagination feature. The previous pagination feature proved very hard to compose with, notably because it reads and writes directly in the dataset and visualizer state rather than using a list of channels directly. Therefore, the pagination interface has been extracted into its own component that uses standard React props and state hooks instead of the overcomplicated and fragile logic and state management system used in the visualizer. The resulting code is simpler and now composes naturally with other features that affect which channels are viewable or not.

Screenshots

For the MPN MEG dataset, with accurate channel information:

For the Face13 EEG dataset, where the BIDS and raw channel names do not match: