Making the ConnectionManager a more "friendly" interface for hooks, resolvers, and template handlers

[jfalkenstein]

The problem

When developing custom hooks, resolvers, and template_handlers, it is often necessary to invoke AWS services. The correct way to do this is via the stack's ConnectionManager so that they can work according to the project and stack configurations for AWS connections (particularly with regard to the profile, region, and iam_role).

I consider the ease of developing custom hooks, resolvers, and template_handlers one of Sceptre's great strengths that distinguishes it from other competing utilities. However, over my time developing these custom extensions, it has not often been straight-forward to invoke AWS services. I've noticed a few ugly bits that I have only been able to make work because of my fairly deep knowledge of the inner workings of Sceptre:

• There isn't really any documentation on the correct way to do this stuff. It's likely that a lot of the custom extensions to Sceptre out there do these things the wrong way. The correct way to use boto3 (with regard to the Stack configs), for example, isn't really clear. We need better docs.
• If you only need to make a simple call to a boto3 service method, you can use self.stack.connection_manager.call() and that works pretty well. It's still not documented, though.
• HOWEVER, if you need to use a boto3 resource object, like an S3 Bucket resource, the only way to do this is via a private, undocumented method _get_session on the ConnectionManager. And to use that private method, you have to know when and how to use, override, or ignore the default ConnectionManager configurations. We need a public interface for getting a session.
• If your hook/resolver/template handler needs to invoke a third-party tool via subprocess, the only way to get authentication to work correctly is to (1) get a session (see the second point above), then (2) use that to get credentials on that session, then (3) create environment variables that can be passed to Subprocess. The means of doing this correctly has taken me a lot of trial and error, isn't documented, and is rather gross. We need an easy way to get authenticated credential environment variables for our Stack Configs to integrate with third party tools.

A few examples of places I've had to do this recently, to demonstrate how "ugly" this process is can be found at:

• https://github.com/Sceptre/sceptre-sam-handler/blob/main/sam_handler/handler.py#L96~L133
• https://github.com/Sceptre/sceptre-cdk-handler/blob/main/sceptre_cdk_handler/cdk_builder.py#L118~L163

How this PR fixes this:

1. I include real docs on how to call AWS services and integrate with third party tools.
2. I've created a public get_session() method that has the "smarts" to normally be used for default purposes with customizability.
3. I've created a new method for creating the environment variables for the session for easy third party integration.

PR Checklist

• Wrote a good commit message & description [see guide below].
• Commit message starts with [Resolve #issue-number].
• Added/Updated unit tests.
• Added/Updated integration tests (if applicable).
• All unit tests (make test) are passing.
• Used the same coding conventions as the rest of the project.
• The new code passes pre-commit validations (pre-commit run --all-files).
• The PR relates to only one subject with a clear title.
  and description in grammatically correct, complete sentences.

Approver/Reviewer Checklist

• Before merge squash related commits.

Other Information

Guide to writing a good commit

[ericabrauer]

LGTM!

[dboitnot]

Looks good.

[jfalkenstein]

@zaro0508 , I've pushed up your requested changes.