Run CI in AWS EC2 instance

[misohu]

1. Change CI to execute in AWS - there is no space in the github runner to run the newly created mlflow image
2. Move the charm code to root folder - MLflow 2.1.1. will be a standalone charm no point to store code in charms subfolder
3. Remove examples folder - as we use 2.1.1. examples will change

The CI uses this action which will create a ec2 instance which was created for the AWS service account (kubeflow-team). All the AWS credentials and GH PAT are part of the secrets (go to settings -> secrets and variables -> actions). All the AWS EC2 configuration is part of the repo variables (go to settings -> secrets and variables -> actions).

CI uses custom AMI created from ubuntu LTS 20.04 AMI and installing docker inside. The AMI is part of the service account private AMI list (its the value set in environment variables).

Publish charm will fail as it is merging agains feature branch not main (but it works on main check this draft pr against main with the same code)

[DnPlas]

Thanks @misohu, few comments.

[ca-scribner]

Don't refactor the current PR, but for a future thought my first impression here is I see both CI changes (aws stuff) and repo structure changes (moving charm from ./charms/mlflowserver to ./). Those sound like good things to separate into different PRs - of 43 files changed, about 30 of them are file moves. (I might have missed a nuance for this case, but ... ) if the repo structure change is a separate PR that only moves the files plus patches the old CI to work with the new file locations (only minimal changes to keep the old functionality), that's a really easy review because we can see the old functionality is maintained. As is, we have to look at those moves and new CI and decide jointly if everything is ok

[ca-scribner]

Some changes. Also, should integration tests be failing during charm build?

[ca-scribner]

(deleted)

Doesn't need to be blocking