3-part solution to OpenData postgres database duplicates

[komasing]

Reason: Postgres database might include duplicates in case of exceedingly long anonymization process or MongoDB failure.

Solution:

1. last_mongodb_timestamp is updated after each successful batch to reduce the amount of records which need to be recalculated in case of MongoDB failure;
2. data lock for Postgres database is introduced - another anonymization process can't commence before the previous process is finished;
3. respective MongoID is stored in Postgres database under mongoid column, which makes it easier to identify duplicates. Note, however, that at any given time it's OK to have 2 Postgres entries with identical MongoID, as producer and client are under different entries in Postgres, yet under one in MongoDB.