The third one of those is mine. It actually has a bunch of communication stuff in it which is completely irrelevant for the epidemic modelling. Further, if you dump the communication elements, you no longer need the extended widgets NetLogo extension. I use a cleaned up version of the model for my ABM training courses, and it is available at http://research.criticalconnections.com.au/ABMBook/. That version does not have communication, but it also does not have population density.
I should also note that the epidemic transmission in the model is actually done with spatial difference equations (so is a system dynamics approach), but the behaviour is modelled with agents.
I ended up actually starting from scratch to build a COVID-19 ABM. It is oriented to social interventions and uses parameters from the literature to set disease state information and is calibrated to a deterministic SEIR model to get reasonable baseline transmission parameters. See https://www.comses.net/codebases/1669ffd5-ed68-4495-96e9-7aa635365ce7/releases/1.0.0/