Performance optimizations.

[benjchristensen]

• memoize dynamic function constructions so executions are statically typed after a single dynamic lookup instead of dynamic on every execution
• use chained observable to reduce composition costs (execute list of functions rather than compositionally invoking onNext)

[benjchristensen]

This pull request is for #104.

I have NOT tried this on production code yet, but all unit tests are working.

I'd like to get code reviews and feedback on the approach before I merge.

[benjchristensen]

NOTE: This is only taking effect on map calls right now. I want to prove it there and then apply to other functions.

It is more or less hacked in on this branch to allow discussion before I do the fairly substantial refactor to apply this pattern elsewhere.

There are also more complicated use cases still to deal with.

[benjchristensen]

I removed the prototype of chaining. This is currently at a clean point that could merge over the dynamic function invocation optimizations.

I am playing more with chaining.

[benjchristensen]

Cleaned up in new pull request with single commit with just function execution optimizations.

#106

Will not merge this pull request.