Add option to gzip compress request bodies

[VimCommando]

When sending large volumes of data to the _bulk API it can be significantly more efficient to compress the bodies of HTTP requests. Thanks to JSON's high compression rates, in my testing I can see over 90% reductions in payload size.

Here are the first and last entires, along with two large payloads, from the same 744,057 document ingest run.

Uncompressed request body, total batch processing runtime: 6 minutes 37 seconds.

@timestamp                  action method status request_path                request_length
Nov 28, 2024 @ 01:21:29.072 bulk   POST      200 /metrics-node-default/_bulk       8,992,884
Nov 28, 2024 @ 01:15:00.898 bulk   POST      200 /metrics-index-default/_bulk     37,895,665
Nov 28, 2024 @ 01:15:00.808 bulk   POST      200 /metrics-index-default/_bulk     38,197,057
Nov 28, 2024 @ 01:14:52.887 bulk   POST      200 /settings-cluster-default/_bulk      42,523

Gzip compressed request body, total ingest runtime: 1 minutes 52 seconds.

@timestamp                  action method status request_path                request_length
Nov 28, 2024 @ 01:13:47.982 bulk   POST      200 /settings-index-default/_bulk     1,072,732
Nov 28, 2024 @ 01:12:00.972 bulk   POST      200 /metrics-index-default/_bulk      1,510,647
Nov 28, 2024 @ 01:12:00.963 bulk   POST      200 /metrics-index-default/_bulk      1,510,034
Nov 28, 2024 @ 01:11:55.512 bulk   POST      200 /settings-cluster-default/_bulk      10,103

These payloads compressed exceptionally well, between 70-95% smaller. Because this was measured from a home ISP to the cloud, the total runtime was significantly more impacted by limited upload bandwidth than the target cluster's compute.

I've added a request_body_compression boolean to the TransportBuilder which allows the request_builder to conditionally compresses the body's bytes and inject the Content-Encoding: gzip header.

[swallez]

LGTM. Thanks!