Stackstorm action runner does not return / complete - hangs and process is defunct

When running an action that makes an internal HTTP call (from inside python with requests) that returns a lot of data. The called service is completing and returning HTTP payload over the network (we wiresharked). The action process (we think ?) (after about 15 seconds after the HTTP server returns its data) goes to "defunct" state. The parent process of the action process is the main action runner process.
The core.http action works for same call ; we cannot use it because it will publish the output to the workflow and that is too much data to be in the workflow context due to Mongo db driver/performance issues.

st2 3.4.1, on Python 3.6.8
Running in Docker & Kubernetes
OS is (cat/etc/os-release):
NAME="CentOS Linux"
VERSION="8"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="8"
PLATFORM_ID="platform:el8"
PRETTY_NAME="CentOS Linux 8"

The report names the action runner, core.http action, Python requests, Docker/Kubernetes, and MongoDB, but no source file or test. Start by reproducing an action that makes an internal HTTP request returning a large payload, then trace the action runner and parent/child process behavior. Done means the action completes without becoming defunct or hanging, while avoiding oversized workflow context.