Clarify the need for mix_stderr in when accessing the output of stderr in tests

[mrchrisadams]

This pull request includes a small change to the docs/tutorial/testing.md file. The change updates the documentation to clarify how to check for output sent to standard error when using CliRunner.

Documentation update:

• docs/tutorial/testing.md: Clarified that the CliRunner instance must be created with the mix_stderr=False argument to access result.stderr in tests.

Why would this change be necessary?

I've made this because I was left scratching my head trying to test the output of stderr today in a project using Typer, and I only found the solution when I came across #385 an issue that was filed in 2022.

I looked for a better exception to raise, but it looks like this more a problem with Click's Result class than Typer; after looking into the codebase there didn't seem to be an obvious place in Typer to create a new, clearer exception class.

So, I've tried to go for some clearer text in the docs instead.

This is the before:

This is the after:

[github-actions]

📝 Docs preview for commit fdbe59e at: https://b8d2cea1.typertiangolo.pages.dev

Modified Pages

• https://b8d2cea1.typertiangolo.pages.dev/tutorial/testing/