[luigi617]

Issue #, if available:

Description of changes:
Run CLI example commands to validate parameter values and fix existing violations

The example command check only parses commands, so values were never checked: invalid JSON, bad shorthand, unknown nested members, values shorter than the model allows, or text passed to blob parameters all passed.

• Added test_example_commands_run to tests/functional/docs/test_examples.py: each API operation and waiter command in an example runs through the CLI driver and stops at before-call, after the CLI has parsed the arguments and botocore has validated the parameters, so nothing is sent to AWS
• Files shown in Contents of blocks are written to a temporary directory and the command's file:// and fileb:// references point at them; each example uses its own files
• Commands run with fake credentials and an empty config. Only argument and parameter validation errors fail the test; errors from the environment, such as a local file the example doesn't show, are skipped
• Custom commands such as s3 cp and emr create-cluster, and Windows examples, still go through the parse-only check only
• Moved the shell parsing of CommandValidator into _command_args, shared by both checks
• Fixed 118 commands in 98 example files, e.g.:
  • \ continuations inside quoted JSON, which bash also passes to the CLI
  • Values shorter than the model's minimum length, e.g. ssm WindowId, kendra index IDs
  • Text passed to blob parameters, now fileb:// or --cli-binary-format raw-in-base64-out
  • Invalid --cli-input-json files, unknown or missing members, wrong types, bad shorthand and --query literals

Testing

• Run pytest tests/functional/docs/test_examples.py -n auto
=== 70161 passed in 53.60s =====
  By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.