Fixes issue #2606 by making color=True behavior agnostic of Operating System.

[Tyl13]

Issue #2606 described a inconsistency in how color=True works between Operating Systems.
This is because ANSI styles can't always be applied to different terminals on windows. This did not cause issues when running it within the terminals because Colorama would change the ANSI style codes to escape codes for windows when it was required. However, if you were to pipe the click.echo() output to a file then the should_strip_ansi function would not work on Windows as the other operating systems.

This is because the auto_wrap_for_ansi function was not being called with the color argument. After the auto_wrap_for_ansi function is called, it defaults color=None and will end up calling should_strip_ansi with color=None which is different from the original call to should_strip_ansi that was called before the wrapper function.

By passing color to auto_wrap_for_ansi, then the behavior of should_strip_ansi becomes consistent.

An additional change applied within the PR is to the tests in test_utils.py. This is to increase test coverage for the different cases of click.echo() and removes the old version of the test. Instead of one test that mocks isatty(), there are now three different tests. One that mocks isatty()=True and _is_jupyter_kernel_output()=False, another that mocks isatty()=False and _is_jupyter_kernel_output()=True , and the last one mocks both isatty()=False and _is_jupyter_kernel_output()=False.

fixes #2606

• Add tests that demonstrate the correct behavior of the change. Tests
  should fail without the change.
• Add or update relevant docs, in the docs folder and in code.
• Add an entry in CHANGES.rst summarizing the change and linking to the issue.

[Tyl13]

Rebased to fix merge conflicts more easily. However, something seems to sometimes cause the windows version to fail. Currently investigating that. I've had it fail locally and then a second rerun succeed. I'm unsure what is causing the issue atm.

[Tyl13]

The issue is coming from my previous change to the runner. The issue is documented here #2732 with the PR to fix it here #2733.

[AndreasBackx]

Thank you, though this was recently fixed by #2607.

[davidism]

Looks like we may have concluded that as well, and now this PR was only test changes? I can't figure out why color was defined as an argument but was never being passed to that function, tried looking through blame history but couldn't figure it out. I'm going to reopen this and see if the tests still merge, as they look like an improvement at first glance.

[davidism]

After rebasing this on latest stable that includes #2607 and #2733, the test_echo_color_flag that you change fails now. I have a feeling the test doesn't need to be changed if the code was changed. I'm not going to block 8.1.8 on this since it's only a change to tests. @Tyl13 if you still remember any discussion about this from the sprint, let me know, and if you can update the test to pass again that would be great.