Retryability for live updates on what seems to be a recoverable error

Expected Behavior

Live update should retry on recoverable errors, or at least be configurable to do so.

Current Behavior

Currently live update doesn't retry on the following error:

Will copy 1 file(s) to container: [<podname>/path]
- 'path/to/file.tsx' --> '/remote/path/to/file.tsx'
Build Failed: Updating pod <podname>: copying changed files: error dialing backend: EOF
LiveUpdate "name:registry.example.com_stuff_here" UpdateFailed: Updating pod <podname>: copying changed files: error dialing backend: EOF

While I'm still able to exec into the pod, get logs from it and interact with it like any other time. Might just be a hickup in my home-network that causes this, so it stings having to rebuild the image for the service whenever this happens.

Context

tilt doctor Output

-> tilt doctor
Tilt: v0.30.4, built 2022-06-16
System: linux-amd64
---
Docker
- Host: unix:///var/run/docker.sock
- Server Version: 20.10.12
- API Version: 1.41
- Builder: 2
---
Kubernetes
- Env: unknown
- Context: rpi8s
- Cluster Name: rpi8s
- Namespace: ---
- Container Runtime: containerd
- Version: v1.23.8+k3s2
- Cluster Local Registry: none

Start by tracing the live-update flow that reports the "copying changed files" and "UpdateFailed" errors, using the shown error output as the entry point. Determine which failures are recoverable and how retry behavior should be configured; done means a recoverable live-update failure retries without requiring an image rebuild, with coverage for the reported EOF case.