Description

This PR introduces support for disabling HTTP keep-alive connections to upstream servers via a new disableKeepAlives configuration option.

Motivation and Context

By default, the HTTP transport maintains keep-alive connections to upstreams for up to 30 seconds. However, this behavior may not be ideal in all scenarios—for instance, when using a Kubernetes service as an upstream, connections can remain pinned to a single pod, preventing effective load balancing. This change enables users to disable keep-alives for upstream connections, offering improved flexibility and control over proxy behavior.

How Has This Been Tested?

1. Set up a Kubernetes environment (e.g., using Rancher Desktop).
2. Deploy multiple replicas of an application (upstream servers) and oauth2-proxy.
3. Configure the Kubernetes service of the application as an "upstream" in oauth2-proxy.
4. With keep-alives enabled, observe that requests are consistently sent to the same pod.
5. Enable the disable-keep-alives option and observe that requests are now distributed across all pods.

Checklist:

• My change requires a change to the documentation or CHANGELOG.
• I have updated the documentation/CHANGELOG accordingly.
• I have created a feature (non-master) branch for my PR.
• I have written tests for my code changes.