Integration of DAE module

[j-koch]

• updates GeneralNetworkedODE class for syntax compatible with DAEs
• new GeneralNetworkedAE class for handling algebraic variable updates
• new integrator - EulerDAE
• New examples to demonstrate DAE, including black-box neural DAEs
• updated NetworkODE examples to conform to new syntax

[drgona]

Failed test:
FAILED test_ode.py::test_random_network - TypeError: SourceSink.init() got an unexpected keyword argument 'state_names'

I suggest the DAE example in the script contains only the DAE part and not the black box NODE.
https://github.com/pnnl/neuromancer/blob/dae_update/examples/ODEs/Part_7_DAE.py
And lets break it into two examples: balck-box DAE and gray-box DAE.

I suggest we create an abstract DAESystem class to serve a similar purpose as ODESystem - no network stuff.
https://github.com/pnnl/neuromancer/blob/master/src/neuromancer/dynamics/ode.py#L53

As of now the API is unnecessarily complicated for the simple example. This is a consequence of working with the GeneralNetworkedODE class.

If designed correctly, we should not need new integrator classes such as EulerDAE.

I believe we should create the following classes:

• ODESystem: base abstract class for ODE systems (black to white box)
• DAESystem: base abstract class for DAE systems (black to white box)
• GeneralNetworkedODE: networked ODE systems
• GeneralNetworkedDAE: networked DAEsystems - i.e., we need to merge GeneralNetworkedODE with GeneralNetworkedAE