The first step in ensuring consistency between Oracle and ES is to obtain
a very basic set of task data - id and timestamp - from Oracle. Add a query
for doing so.

es.get() raises NotFoundError in both cases - when index does not exist and
when document does not exist. Also, it's more reasonable to check index once
since it's the same for all messages.

- Add/update functions and their parameters' descriptions.
- Update the script's description.
- Add consistency' description into README.

Check that all fields supplied in input data are present in ES and their
values are matching the input data, instead of working only with tasks and
their timestamps. This will allow checking tasks' other fields as well as
different types of documents such as datasets.

Add stage 016 into consistency chain because it adds the fields required
for getting documents of given type from ES.

Prepare the script for further development, where incosistent tasks will be
automatically reloaded into ES.

These functions are either used by several scripts or will be in the future.
Move them to library to uphold DRY principle.

DEBUG mode in data4es-start exists to check the workflow without uploading
anything to ES. Consistency check writes nothing, so DEBUG is unnecessary
here.
Do not redirect the stages' stderrs, leave them as-is.

While the script is the stage 069's counterpart in data4es-consistency-check,
they share no functionality.

- State what is retrieved by the query.
- Remove unnecessary information.

- Show an error message and exit if no host, port, or index is specified.
- Remove default values of the parameters.

Printing all discovered inconsistent records to stdout as a batch contradicts
with various things, such as pyDKB's file mode and the possibility of
controlling the workflow with Apache Kafka.

Create an output message with _id and _type for each inconsistent record.
Still exit with code 1 if at least one inconsistent record was found,
0 otherwise.

- Add output samples for stages 009 and 016. Stage 071 has no sample because
  all the tasks in 016's sample are up to date in ES.
- Update READMEs.
- Add symbolic link for stage 071's input.

Add information about the new stage and dataflow to the README.

Type of _id is unknown - it can be str or int for task, and str for dataset.

The field is required to get child documents such as output datasets.

Service fields are different from data fields and shouldn't be checked.

These are unnecessary because library files are not supposed to be executed.