Support Query RPC for fast first page results

[scotthart]

BigQuery supports both the InsertJob RPC (with type QueryJob) as well as a Query RPC (https://github.com/googleapis/googleapis/blob/8798ceff3f6fbcdce3186b67ce9339df337569d5/google/cloud/bigquery/v2/job.proto#L117). InsertJob is an asynchronous RPC stores the query results in a BigQuery table, while Query is a synchronous RPC that returns the first page of results immediately while executing a QueryJob asynchronous for the rest of the results.

Add a Query method that returns the first page of results as a collection of std::tuples and a JobReference that can be used to poll for the remainder of the results.

[cuiy0006]

while executing a QueryJob asynchronous for the rest of the results.

Is this done by GetQueryResults?

Can GetQueryResults be executed in parallel or it must be executed in sequence? Because I see there is page_token needed to be passed in the request of the next page, it is returned by the response of requesting the last page.

[scotthart]

GetQueryResults is available in https://github.com/googleapis/google-cloud-cpp/blob/6055c7025d1c98625fbcbbe7a7eb4a74577a1c9b/google/cloud/bigquerycontrol/v2/job_client.h#L291 but there are no current plans to support it in this library.

GetQueryResults transfers data as JSON via a paginated mechanism over REST. The ReadArrowmethod is a streaming read that can be parallelized offering superior throughput.