Add protobuf codec for query range and instant query responses

[afhassan]

What this PR does:

This PR introduces a protobuf codec for encoding query range and instant query responses returned from Querier to Query Frontend (QFE). Protobuf encoding significantly reduces the latency between Querier and QFE for large size query response as it is much faster than json encoding. The new protobuf codec converts PromQL responses to a PrometheusResponse and then marshals it to Protobuf. Once the marshalled response is returned to QFE it is deserialized and merged with other responses if the query was sharded.

The PR also introduces a new config to specify the response compression. Currently the supported values are "" (no compression) or "gzip"

For unsharded instant-query requests, a JSON response is returned.

The impact on an unsharded range query with 1M samples is shown in the images below:
{__name__=~'metric_([0-9]|[1-9][0-9])'}

Before:

After:

Which issue(s) this PR fixes:
Fixes #

Checklist

• Tests updated
• Documentation added
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[justinjung04]

@afhassan could you fix the DCO that's failing?

[yeya24]

Revisiting this PR, it looks like with the current version of Prometheus, we don't need a dedicated Prometheus handler anymore.
You can implement a new Codec for protobuf and use it in Prometheus API. https://github.com/prometheus/prometheus/blob/main/web/api/v1/api.go#L284

[yeya24]

Looks good! Can we update the PR title to make it more descriptive so that reviewers can understand the change? Also the PR description is outdated.

Let's also add a changelog entry

[yeya24]

https://cortexmetrics.io/docs/configuration/v1guarantees/#experimental-features

Sorry for the back and forth. Can we also mention it as an experimental feature?