migration to spark 3 (#5)

Merge pull request #4 from adidas/release/5.3.0

introduction of new algorithms (including DeltaLakeLoad using Delta Lake IO); refactored project structured; full load performance improvements; bug fixes;

Merge pull request #3 from bemu/develop

do not compute statistics separately as this is handled by the TableStatistics trait

* use fail-fast for full loads as default read mode

Merge pull request #2 from bemu/develop

* remove credential id and label

* add algorithm for flattening and exploding nested fields

Merge pull request #1 from adidas/semistructured_data_load

Add support for semistrucured data loads

initial commit