Split connector code into streaming and batch

The current codebase is overloaded with conditionals, interconnected or ignored properties and doesn't take into account different limits of [BigQuery API](https://cloud.google.com/bigquery/quotas#load_jobs). That makes it harder to support and may lead to unobvious configuration problems.

The proposal is to split the current monolithic sink connector into separate artifacts, one for streaming via Storage Write API, another through GCS load jobs. It would make the code much easier to understand and support.