v0.4.0

v0.4.0: Synchronization performance improvements

A number of changes to synchronization that improve speed (up to 2x) and
memory usage.

• Instead of loading all observations and then processing them, they're
  downloaded 100 at a time and processed as they're downloaded. For
  people with a large number of observations this can drastically
  reduce memory used during full sync.
• Instead of downloading all observations and then processing them,
  processing of a batch begins while downloading the next batch.
• Change JSON library to dkjson for an improvement in JSON parsing
  speed. On my machine this reduces JSON parsing time from about 1.5s
  per 100 observations to 0.7s. Code size is also reduced.
• Improvements in updates logging for easier log analysis.
• Fixed logger initialization so that plugin reload is not required to
  start logging.