bpo-33522: Enable CI builds on Visual Studio Team Services

[zooba]

We have been provided with https://python.visualstudio.com/cpython and a measure of build machines to help us run continuous integration builds. The benefits are more reliable builds (especially Mac and Windows) and higher parallelism than our current services.

This PR adds the build definition files and fixes some test bugs to ensure we can successfully test on this platform. There are a few more manual steps to make it work against our repo, which we will do immediately after merging the changes.

There are also some more changes coming, but feel free to start the review :)

https://bugs.python.org/issue33522

[miss-islington]

Thanks @zooba for the PR 🌮🎉.. I'm working now to backport this PR to: 3.6, 3.7.
🐍🍒⛏🤖 I'm not a witch! I'm not a witch!

[bedevere-bot]

@zooba: Please replace # with GH- in the commit message next time. Thanks!

[bedevere-bot]

GH-6923 is a backport of this pull request to the 3.7 branch.

[miss-islington]

Sorry, @zooba, I could not cleanly backport this to 3.6 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker e5f41d2f1e0b8b8e61d5fa427c19bd1ea90fd9a3 3.6

[bedevere-bot]

GH-6925 is a backport of this pull request to the 3.6 branch.

[miss-islington]

Thanks @zooba for the PR 🌮🎉.. I'm working now to backport this PR to: 3.7.
🐍🍒⛏🤖

[bedevere-bot]

GH-6926 is a backport of this pull request to the 3.7 branch.

[gpshead]

The test and standard library changes should not have been part of this PR. Now those have been committed, including backports to two stable releases, with no description of what changed in the commit message and inadequate news entries.

The Library NEWS entry fails to mention that the change was Windows specific.

The test.support Library change isn't even documented with a NEWS entry.

There is no description of what changed at all in the commit message. The commit message is the most important place to record such information. It is the only thing we have on permanent record. This PR and Github are fly-by-night transient sources of information that can and will disappear in the future.

@ned-deily

[gpshead]

@ned-deily

[ned-deily]

There are a lot of changes in this PR and, AFAICT, there are changes to non-Windows tests. Do we know if any of the changes correct problems already documented in other b.p.o issues? I agree with @gpshead that such a pervasive change, especially coming in at the last minute, should be better documented (PRs are not sufficient). Downstream users want to know what has changed between, say, 3.7.0b4 and 3.7.0rc1 without having to resort to exhaustive diffs; that why we have NEWS entries and the changelog html.