Implement data- and platform-dependent estimates of calculation times

Getting an ETA on queued calculations would be highly informative to users.  Such ETAs depend on data sets, platform used, and queue depth.  Implement tracking of time via attributes (depends on the attributes issue).  Develop an API for getting platform- and method-specific scaling from subdividing a longish test set.  