SPARK ODBC : overwrite if not exist and append existing

[saubert]

Spark ODBC is really faster than Hive, at least for GUI.

However, two things are missing :
1/ Append existing for the write date (exists his way on Hive)

2/ability to put "overwrite" even if the table does not exist (it works this way on Hive)

These two drawbacks limit severly th

[saubert]

It appears that the overwriting of a non-existent table didn't generate any error in 10.6 on Hive... but generates also an error in 2018.3. This is big regression we will have to investigate with the support.