Update for MaxWell systems

[cmccrimmon]

MaxWell channel-electrode mappings may not be bijective (see #1703). Throw a warning rather than an error if this is the case and address this correctly downstream in spikeinterface modules.

[samuelgarcia]

Hi.
Unfortunatly, this is not possible, we need to ensure that the ids are unique.
What we can do is to use the channel name in maxwell (which can be none unique).
And also we could have a hook in maxwell channel_ids generation a check of unicity and in case it is not unique make a transformation of the id to make it unique (like adding the index of the channel)
What do you think ?

[h-mayorquin]

Using names should be OK, right? Then people can use the "use_names_as_ids" in SpikeInterface

https://github.com/SpikeInterface/spikeinterface/blob/fb72ed2af963e257a2a12b1d4280f622c259c7c3/src/spikeinterface/extractors/neoextractors/maxwell.py#L33-L35

They names are also propagated as a property that can be used in SpikeInterface to fix things downstream.

[cmccrimmon]

Using names should be OK, right? Then people can use the "use_names_as_ids" in SpikeInterface

https://github.com/SpikeInterface/spikeinterface/blob/fb72ed2af963e257a2a12b1d4280f622c259c7c3/src/spikeinterface/extractors/neoextractors/maxwell.py#L33-L35

They names are also propagated as a property that can be used in SpikeInterface to fix things downstream.

Yes, using names should be fine as these are propagated and can be subsequently dealt in SpikeInterface. A singular enforcement of uniqueness should be implemented here then (throw out any duplicated names after the first) as required.

[zm711]

We discussed this in our meeting and we can't have this change in Neo-baserawio. We will need to fix this at the Maxwell level. Feel free to open an issue to discuss this more.