Epydemic

- Fixed behaviour for processes with no loci
- Added type annotations
- Re-configured to use epyc >= 1.0.1 API

- Refactored to add network generators
- Removed tutorial on building networks at experiments, since generators solve that
- Added generators for the common network science network classes
- Added tutorial on building a new network generator
- Cross-referenced a lot of the documentation
- Added notes to tutorials where they seemed unclear
- Renamed AddDelete.remove() event to delete()
- Added cookbook recipe on dynamic population simulations
- Added discussion of clustered networks to cookbook recipe on human contact networks
- Added more references to bibliography

- Removed use of six, since we're no longer supporting Python 2.7
- Removed six and future from requirements
- Replaced standard Python random number generation with the new numpy
generator class
- Added SEIR model
- Documented SIRS model
- Slightly re-structured the presentation of the API documentation

- Added superclass calls in models that were preventing multiple inheritance from
working properly
- Refactored events and loci into Dynamics rather than Process
- Defined a monitor process to capture the sizes of all loci as time series
- Changed names of parameters and results to be more structured, to avoid conflicts
- Added some implementation notes

- Removed Python2.7 compatibility, since NetworkX v.2.4 doesn't support it anyway
- Fixed compartmented loci for adaptive networks
- Adding an edge where the endpoint nodes don't exist now throws an exception rather
then creating the nodes automatically, since there's no sensible general default
behaviour
- Removed the maintenance of nodes in different compartments, since the
same effect can be had just as easily (and with less space) using loci

- Started maintaining lists of nodes in different compartments to
remove a bottleneck for adaptive networks