Abstraction for manipulating LTI systems

[arvoelke]

Create a class that accepts a system in any number of forms:

• state-space
• transfer function
• zero-pole
• nengo.synapses.LinearFilter

Then functions such as impulse, is_exp_stable, scale_state, and state_norm must accept these objects. The objects may also support multiplication, operation, and comparisons.

[arvoelke]

Also might as well make it handle MIMO, thereby addressing scipy/scipy#5753

[arvoelke]

Basic functionality in 641662c, supports:

• state-space
• SISO transfer function
• addition
• multiplication
• powers
• inversion

Currently reconsidering the challenge of handling MIMO. One issue is that multiplication is no longer associative. Another issue is it becomes difficult to disambiguate the many possible representations for the system. In particular, if the shape of the array has two dimensions, it could be SIMO, MISO, or even MIMO with a mismatch in the lengths of the numerator polynomials. Finally all the use cases and examples are currently SISO and so I'm afraid this just introduces unnecessary complexity (at least for now).