Better support for specifying state variables of composite machines

[slibkind]

Following this discussion, it is clear that the current system places the onus of proper bookkeeping of the state of a composite machine on the user. A new feature that allows the user to use the wiring diagram to specify a state as a composite would reduce some of this burden.

There is also interest in having "stateful" machines, i.e. those which are equipped with a current state. This may have performance implications.

[slwu89]

link #88

[slibkind]

For delay differential equations also specify history functions as part of the machine