Fix bug where later causes 100% CPU in R from terminal

[jcheng5]

Fixes #74

Testing notes

See the repro example in #74, and also rstudio/websocket#34. It must be tested on Mac and Linux; Windows is not affected.

[jcheng5]

Switch *callbackRegistry. to callbackRegistry->

[wch]

Minor nit: Replace *callbackRegistry.nextTimestamp() with callbackRegistry->nextTimestamp() to make it a bit more obvious what's going on.

[jcheng5]

@wch It turns out callbackRegistry is not a pointer, but rather, nextTimestamp() returns an Optional<Timestamp> which you dereference to get the actual Timestamp value. I've pushed a commit to make this a little more obvious.

[wch]

FWIW, with boost::optional, I find it clearer to use callbackRegistry.nextTimestamp().get() instead of *(callbackRegistry.nextTimestamp()), precisely because of the confusion we're having here. (We're using the local Optional class instead of boost::optional, though.)