Refactor envoy configuration to using upstream yaml.file

[orfeas-k]

Context

We should refactor the charm to be configured using a simple yaml file instead of generating it using the envoy_data_plane package for the following reasons.

1. During the update of envoy for Kubeflow 1.9, there were updates to the envoy.yaml file that are visible in PR fix(metadata envoy): upgrade envoy and config from 1.12 to 1.27 kubeflow/pipelines#10589. Regarding those:

• Some added typed_config fields are not supported at the moment from the specific package.
• Envoy version had a major bump (1.12 -> 1.27) which means that we may need to bump the package in order to support new configurations. That being said, the package is currently pinned and bumping results in errors when packing the charm.

2. In general, it is a more cumbersome developer experience to map changes in the envoy.yaml file to the correspond. In other words, it's easier to just copy changes from upstream if we also use a yaml file.
3. This has caused us trouble in the past e.g. in this PR fix: Update envoy configuration #64 (comment) we had to revert changes because the python package didn't work the same way with the yaml file (see bullet about revert max_grpc_timeout). This also resulted in us having a slightly different configuration than upstream.

Thus, in order to make our lifes easier, I think we should refactor the

What needs to get done

Replace the config generator component with a simple yaml template being rendered and pushed in the container.

Definition of Done

Envoy is configurable through a yaml file.

[syncronize-issues-to-jira]

Thank you for reporting us your feedback!

The internal ticket has been created: https://warthogs.atlassian.net/browse/KF-5742.

This message was autogenerated