drop_late_spikes as a Population param and vs Delays

[Christian-B]

We have a cfg setting:

[Simulation] drop_late_spikes = True

But the user could also pass drop_late_spikes in as a Population param This does not appear to be standard PyNN

Is there actually a use case for allowing users to set drop_late_spikes on a per Population basis? or do we keep just the cfg setting?

My Proposal would be to drop the Population param.

Note: This would also affect ExternalDeviceLifControlVertex However there is no known use case of reading it out of such a vertex So as we don't appear to tell the external device what to do setting drop_late_spikes should not affect the external device anyway

=== I also notice that DelayExtensionVertex only uses the cfg

Now ideally the Delay should have the same drop policy as its target. But what if there are two targets. drop if all drop would be the only thing that makes sense.

==== This was discovered as part of https://github.com/SpiNNakerManchester/sPyNNaker/issues/1180

Although for the DelayExtensionMachineVertex that can easily just read the cfg as that is all the Application Vertex does now anyway

[rowleya]

I suspect drop_late_spikes as a Population parameter is non-critical; it is only really used at the simulation-level as far as I know.