add spike source docs.

[thorstenhater]

Explain where spikes come from.

Closes #1647

[Helveg]

Is there anything reasonable that can be done in the code example to show "oh hey, we have 3 spike detectors, so now we have spikes with id 0-2", that'll show the input output relationship?

[thorstenhater]

Yes, @brenthuisman earmarked this for himself. But the gist is, we are going to attach a metadata field to the spike callback, eg location as s-exp string and label string, so you can engage in some filtering. At least in C++ that should work, for Python performance might be an issue.

Although this conflates issues, see here: #1929

[thorstenhater]

Fixed: #1941

[thorstenhater]

Also, there is a stub in /doc/concepts/decor.rst for this. Maybe move this documentation there, since talking specifically about being placed (should that list of placeables be termified then?). A second (more abstract?) definition of a spike detector might be confusing.

That's the core issue right there, they are both part of the data collection framework and of the networking facilities.
So, where to categorise them? I'd prefer to explain the detector as part of the interconnectivity as that is the topic where
I consider them most important. And then -- of course -- link to data extraction (you know, you can actually write out
those spikes we use to network) and how to set them up (since that's how they come into being).

[brenthuisman]

That's the core issue right there, they are both part of the data collection framework and of the networking facilities. So, where to categorise them?

=> they should be in both places. By adding a link I'll let you get away with adding text just once :) BUT, it would be confusing to have some of them in decor->placeables, and others under interconnectivity. Currently in the former there lives 1. Connection sites and 3. Gap junction connection sites. Could we then move that over too (and replace with links)? And maybe we can add a few lines (to decor->placeables) on the subject of spike detectors that explains their dual use and details their use for the case where people want to get spiking times/counts out?

[brenthuisman]

I now approve of this message.

[thorstenhater]

I likewise approve your additions ... it's still funny to me that we can circumevent review process this way.