The datasets created here contain non-specific data (a floating point number)
along with an record identifier. The created data is designed to produce six
distinct clusters, suitable for K Means Clustering. Paired datasets are created
containing these clustering pairs along with a significant number of non-paired
random data values.  These datasets are named according to their owners ("alice"
and "bob"), so the data looks something like this:

  Alice's portion of a PSI Vertically Partitioned dataset for training
    data_0,     ID0
    -9.48380,   88888
    17.1271,    22222
    9.1563,     33333

  Bob's portion of a PSI Vertically Partitioned dataset for training
    data_1,     ID1
    1.6878,     66666
    2.08873,    88888
    -10.3749,   22222

In this example, two records (88888 and 22222) have common identifiers across
the datasets and can be used for clustering analysis. The records for 33333 and
66666 will be ignored. This "vertically partitioned data" will later be used in
examples as if the data was in one place like this:

  Logical view of overlapping data
    data_0,     data_1,     ID
    17.1271,    -10.3749,   22222
    -9.48380,   2.08873,    88888


The generated data is stored as prepositioned assets for use in examples
within this SDK.  These created assets are named:
 EXAMPLE - alice's portion of a PSI Vertically Partitioned dataset for training.
 EXAMPLE - bob's portion of a PSI Vertically Partitioned dataset for training.
 EXAMPLE - alice's portion of a PSI Vertically Partitioned dataset for inference.
 EXAMPLE - bob's portion of a PSI Vertically Partitioned dataset for inference.
