Speed up bulk processor flush and cut release

[philkra]

• improve through put of flushing the bulk processor
• set release artefacts

[kostasb]

Consistency and speed issues are resolved.

The queue's stats are slightly off by one or two batches.

Example, for a dataset with 10k records and default parameters:

bp.flush_queue()
print(bp.stats)

Output:
{'total': 9975, 'queue': 0, 'failed_batches': 0, 'total_batches': 399, 'tables': {'test': 9975}}

[kostasb]

Tested and works well.