adding support for the multi-query processing (both candidate generation and refinement); a few bug-fixes

Version3

Approved-by: David Novak <novak.david@gmail.com>

moving all answers to a different package; changing hierarchy of operations and answers; answer 'ProcessObjectsAnswer' now contains 'skipped_objects' field with reasons to skip the objects

each CRUD answer now contains field 'skipped_objects' with a list of objects that were not processed; each of these objects has a field '_reason' with the reason it was skipped'

fixing the way we work with statuses, especially when merging them; now calling Algorithm.init() in AlgorithmRestore;

statistics now work like this: 1) one must turn them on/off using the global setter (like before), 2) one can configure the algorithm to bind statistics for each operation (local thread stats) - use OperationLogger.java  class, 3) if you want distance computations to be evaluated automatically, use StatisticDistanceFunc wrapper

added HTTP client and record processor to use this HTTP client; proper usage of HTTP status in exceptions; AbstractAnswer now implements in a default way wasSuccessful() method

ErrorCode renamed to ProcessingStatus; OperationEC and BucketEC removed and corrsponding statuses were moved to individual operation answers; HttpApplication now properly processes excpetions and returns an error answer with the stack trace in the text

added a proper processing of PrintAllObjects operation - it writes output to a tmp file and then prints to JSON (and deletes); operation logger writes operation + answer to JSON just up to a certain level (parametrizable); underlying record for operations/answers is now always modifiable