New EMG examples showcasing invasive recordings

[klotz-t]

The current EMG-BIDS specification does not include specifications for invasive EMG recordings (which are also highly relevant to the field). This pull request contains three examples of invasive EMG: concentric needle, fine wire, and combined invasive and surface EMG.

The following extensions (in _emg.json) should suffice to make EMG-BIDS capable of handling invasive EMG:

• EMGType (REQUIRED): Specify the type of EMG recording. Can be either surface, invasive, or combined.
• CannulaDiameter (RECOMMENDED if invasive EMG): Diameter of the inserted needle (it is often reported in Gauge units; decide if one wants to stay with that convention or if metric units should be used (more BIDS-like))
• CannulaLength (RECOMMENDED if invasive EMG): Length of the inserted needle in millimeters.
• CannulaMaterial (RECOMMENDED if invasive EMG): Material of the needle.
• InsulatorMaterial (RECOMMENDED if fine wire or concentric needle electrode): Material used to insulate a pair of fine wire electrodes or the signal electrode from the shaft (reference) in a concentric needle.
• SterilizationMethod (RECOMMENDED if invasive EMG): Method used to sterilize the EMG-electrodes, e.g., E-beam or ethylene oxide.
• ElectrodeSurfaceArea (RECOMMENDED if invasive EMG, otherwise also not a bad idea to be reported): Surface area of the electrode contact in mm^2 (in the case of fine wire electrodes, only fully specified through ElectrodeDiameter (does also fit other electrode types) and ElectrodeTipLength (potentially only relevant for that type of electrode))

Here are a few additional considerations from validating these new examples with the BIDS validator (all of them pass):

• As it is, there are fields such as ElectrodeManufacturer or ElectrodeManufacturersModelName that are identified to be included in *electrodes.tsv (if needed); however, they are not specified there, so the BIDS validator only lets you pass if you put these fields in an electrode.json file.
• InterelectrodeDistance is currently described to belong to *electrodes.tsv (if heterogeneous); however, formally it is defined to belong to *channels.tsv. I think it makes sense to have it in the electrodes.tsv; at least, it should be described consistently.
• It can be kind of tricky that SoftwareFilters is a required field; if channel-specific, there is nothing reasonable to put there (n/a also feels not correct). Maybe there should be an option channel-specific (next to n/a or object). In one example, I have implemented a workaround: in the corresponding object, I set the low- and high-pass cut-offs to channel-specific (to let you pass the validator).