Have a clean client-server interface for muscle sim

[Samir Menon]

It's about time. The existing muscle setup is clunky and makes it really annoying to deal with simple analysis..

[Samir Menon]

We have a separated vis : 1ed7cf614c8fde5701f0596073fd0a4d161adc5e

Now we're left with dynamics and control. We'll adopt a strategy based on whether the research questions involve segregating muscles from dynamcs and control.

So we have two scenarios: 1. When the muscles simply play the role of actuators. Aka., we'd like to study actuator kinematics while using the physics and control as tools to jiggle the model. Could also be used when we merge muscle inertia into rigid body inertia (this is obviously the case in the standard model), and when we develop muscle-structure agnostic controllers (which we do when we assume that muscle projections don't shape task forces).

This is readily handled by appending muscles to an existing control/dynamics sim. Enough for answering quite a few research questions. So we'll focus on this first.

2. When muscle inertias and force properties play a role in shaping dynamics and control directly. This is obviously a better model and we should converge to it eventually. However, it will require more complex programs and perhaps we'd best avoid it for now.

Samir.

[Samir Menon]

Increasing priority.

[Samir Menon]

This is mostly done. It's too clunky to have separate apps for control and muscles.

So we'll just have one muscle control app. Might duplicate some code. But that should be resolved by moving init stuff to init.