Fixes #3658

Summary/Motivation:

The recent rework of capture_output to make it more robust (#3583, #3588, #3601, #3633, #3640) led to a situation where using capture_output(capture_fd=True) on Windows would cause the process to deadlock (see #3658). We tracked it down to what appears to be Window's procedure for reallocating (growing) the buffer that underlies os.pipe(). This PR proposes a two-fold solution:

1. Request a larger pipe buffer (64kb), to reduce the chances that the OS will ever need to reallocate the buffer
2. Mark the pipe as being non-blocking (PIPE_NOWAIT), on the theory that losing log data is better than deadlocking the process

Changes proposed in this PR:

• Provide windows-specific implementation of os.pipe() to request a larger buffer
• Make the pipe underlying capture_output non-blocking

Legal Acknowledgement

By contributing to this software project, I have read the contribution guide and agree to the following terms and conditions for my contribution:

1. I agree my contributions are submitted under the BSD license.
2. I represent I am authorized to make the contributions and grant the license. If my employer has rights to intellectual property that includes these contributions, I represent that I have received permission to make contributions and grant the required license on behalf of that employer.