services: make the AWS SDK configurations more obvious and easier to work with

## Problem
The SDKs pull in configuration data from a lot of places: environment variables, shared INI files, client constructors (within the Toolkit), etc.

Developers should be able to easily figure out what state the SDK/Toolkit is in when making API calls. An environment variable like `AWS_USE_DUALSTACK_ENDPOINT` can cause the Toolkit to fail in mysterious ways with little explanation as to _why_ things are failing. Also, exposing more of the internal configurations could eventually allow users to do things like define their own endpoints.

Start by tracing the configuration sources named in the issue: environment variables, shared INI files, and Toolkit client constructors. Determine how the SDK and Toolkit state can be inspected and how configuration failures should be explained. Done means the configuration is more visible and the issue's endpoint-customization goal has a defined, tested design.

Written by the indexing model from the issue text.