fix: enhance kakfa windowing for streams and enable avro kafka message format

Major fixes:

Make windowBy to be not just a column name but a spark sql expression. This allows for more advanced selection of custom columns for windowing
Dataframe utils to prepare window and event timestamp columns are updated accordingly.
Streamable virtual sources now allows to redefine windowBy column.
Enable Avro format for kafka key and value message columns.

Minor fixes:

Refactor dumpSize defaults for error collection: when dumpSize is missing in target configuration then application level dumpSize value is used.
Weaken configuration validation requirements for spark parameter strings (enable string with various separators including new lines).
Add debug logging for schemas (both during schema reading and during sources reading).
Fix merge strategy rules for assembling dependencies uber-jar
documentation updates to reflect aforementioned changes.

Raiffeisen-DGTL / checkita-data-quality