Implement symlinks in Marquez

[pawel-big-lebowski]

Problem:

We need an ability to store alternative dataset names. For example hive datasets can be identified by their data files` location or metastore uri with database and table.

Solution in Spec:

SymlinksDatasetFacet -> OpenLineage/OpenLineage#936

Implementation in Marquez:

Model changes:

• Create extra dataset_symlink table in Marquez with columns: (symlinkUid, name, namespaceUid, symlinkType)
• Replace name field in datasets table with symlinkUid

Implementation follows the proposed DB changes:

First PR -> reflect current behaviour in modifed schema

• provide migration SQL for existing instances
• create a dataset_symlink row whenever dataset is created
• modify SQLs in dataset_version_dao, etc.

Second PR

• Extract symlink facet when posting new OpenLineage event
• fill dataset_symlink with multiple entries per OL event.