'reduced' test simulation duration vs. `tstop`

[cjayb]

In #228 we found that truncated event times due to tstop cause discrepancy between desired and actual drive spike trains A new test was added in #221 (test_dipole.py) to assert these are identical. This required changing the 'reduced' network simulation in _run_hnn_core_fixture from 25 to 40 ms.

See also #230

[jasmainak]

what is the action point here?

[cjayb]

To decide whether the additional 15 ms of simulation time makes the tests run too slow. I don't think so, but this could also become a moot point if #230 gets fixed.

[blakecaldwell]

This is relevant to optimization in HNN GUI which stops a simulation early (reduces tstop) if it doesn't need to calculate RMSE over that time region. Should tests be run for 15 ms longer than necessary in order to consistent results for the desired window?

[ntolley]

Looks like this is no longer an issue since #230 is merged