Parallelize tests

…ghlight its positivity

…o highlight its positivity

… more readable

Check if replication went correctly after each query

Cleanup datetimes

Replicator consumer wrongly assumed that the values for original
table's clustering key coming from CDC Log table are not null.
However for some queries like range deletes clustering key may
be unspecified. An example of such a query is:

SELECT FROM table WHERE pk = 0;

This would result in two rows in CDC Log table both having null
as a clustering key value.

As a remedy, types for clustering key values and in consequence
all the other columns' values have been changed from &CqlValue
to Option<&CqlValue>.

…en batch is not finished instead of false

…the last read row should not be skipped

…n the loop

…tinuously

Fix fetch_generations_continuously

Implement range row deletes

…plication

…emTime to chrono to make it more consistent with the rest of the library