Refactor request execution to use Execute() on request objects

[danxmoran]

Our codegen supports executing constructed requests in one of two ways:

1. Clients have a <OperationId>Execute method per operation
2. Requests have an Execute method that delegates to the appropriate ^^^ method under-the-hood

Initial commands have used style 1, but after seeing style 2 used in #52 and #53 I think it's a better approach, because:

• It allows for building and executing requests in a single statement, reducing noise
• In the future, it reduces the amount of context we need to pass around in order to extend & execute partially-constructed requests

Refactor the few commands that are already in place to use this style.