Connection between bundle manager and worker is flaky

Created by: efritz

A customer has recently showed us that with LSIF uploads between 1-2G on a k8s deployment will see failures 50% of the time from the worker due to a connection reset when trying to fetch the raw upload data.

We need to investigate the connection between these two services to determine whether it's a timeout issue or something else. It may be necessary to split the payload into smaller chunks as we do for uploads.