Closes #1118
Closes #1165

Co-authored-by: Sergey Puchnin <spuchnin@gridgain.com>
Co-authored-by: Taras Ledkov <tledkov@gridgain.com>
Co-authored-by: Oleg Ostanin <oostanin@gridagin.com>

* Adding MapR Binary & JSON YCSB Clients
* MapR binaries will be picked from the classpath

[accumulo, accumulo1.6, accumulo1.7, accumulo1.8] switch Accumulo to minor version specific bindings.

* break out accumulo modules
* add deprecation warning
* switch travis to use short hostname to get around long hostname error.

* remove mapkeeper and voldemort from command help, since we don't build them
* fix ordering of modules in root pom

closes #328

the existing YCSB API to generate time series data (numeric values associated
with timestamps).
See the TimeSeriesWorkload.java class Javadocs for details on how to use it
with a client and workloads/tsworkload_template for parameters.
Also add the BasicTSDB instance for printing/debugging the workloads.

This commit is an initial cut at a complete implementation using the
low-level Elasticsearch REST client.

[hbase12] Add HBase 1.2+ specific client that relies on the shaded client artifact provided by those versions. (#970)

[arangodb] Add ArangoDB Support

[scripts] Add mvn build if necessary for source checkout on windows.  Plus a couple other fixups to make consistent with the linux script.

Note the Guava classes in the test directory for AsyncHBase. These are there
as AsyncHBase uses Guava >= 18 while HBase and the mini cluster use Guava 12.

[riak] Done some code/comments cleanup.

the native Protobuf GRPC based API.

Added new binding for Riak, based on the one created by Basho Technologies (available at https://github.com/basho-labs/YCSB/tree/master/riak).

Change made to the original version:

- Made compatible with the latest YCSB version.
- Removed unused Constants.java file.
- Added strong consistency test support.
- Done some refinements to the code.
- Removed many unused function in the RiakUtils.java file.
- Choosed to use executeAsync methods to resolve a bug that didn't allow the benchmark to end whenever one or more nodes of the cluster are killed.
- Added TIME_OUT status return code.

To do:
- Add different LOG errors (i.e. insufficient_vnodes_available, ring_not_ready, etc.) - for references, check page http://docs.basho.com/riak/latest/ops/running/recovery/errors/#Errors-and-Messages
- ...