[DS] Generate repositories partitions from HDFS blocks

[ajnavarro]

Repositories will be in a specific folder. Example:

repositories/
├── repo1/
│   └── ...
├── repo2/
│   └── ...
└── repo3/
    └── ...

We should get all the files from a repository, get the blocks information, and aggregate repositories by datanodes with more block of each repository.
With this information we need to create a new class called RepositoryPartition that extends the trait org.apache.spark.Partition, that will include a list of repository folders.

This partitions will be sent to each relation to create RDD partitions correctly, depending of the locality.

[ajnavarro]

not needed anymore.