When handling \s, \d, or \w (and their inverse) escapes in bytes regexes this a small but measurable performance improvement.

<!-- issue-number: [bpo-34636](https://www.bugs.python.org/issue34636) -->
https://bugs.python.org/issue34636
<!-- /issue-number -->

Followup to 90fc8980.

<!--
Thanks for your contribution!
Please read this comment in its entirety. It's quite important.

# Pull Request title

It should be in the following format:

```
bpo-NNNN: Summary of the changes made
```

Where: bpo-NNNN refers to the issue number in the https://bugs.python.org.

Most PRs will require an issue number. Trivial changes, like fixing a typo, do not need an issue.

# Backport Pull Request title

If this is a backport PR (PR made against branches other than `master`),
please ensure that the PR title is in the following format:

```
[X.Y] <title from the original PR> (GH-NNNN)
```

Where: [X.Y] is the branch name, e.g. [3.6].

GH-NNNN refers to the PR number from `master`.

-->

This reverts commit 886483e2.

GCC complains:

Python/pylifecycle.c: In function ‘_Py_InitializeFromConfig’:
Python/pylifecycle.c:900:13: warning: ‘interp’ may be used uninitialized in this function [-Wmaybe-uninitialized]
         err = _Py_InitializeMainInterpreter(interp, &main_config);
         ~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

This seems spurious since &interp is passed to _Py_InitializeCore. Anyway, we
can easily initialize to quiet the warning.

closes bpo-31902: Fix the col_offset attribute for ast.Async* nodes to point to the "async" keyword. (GH-4175)

Previously, col_offset points to the keyword after "async".

Python 3.x does not fall back to comparing object addresses when comparing two `dt` objects.

<!-- issue-number: [bpo-34365](https://www.bugs.python.org/issue34365) -->
https://bugs.python.org/issue34365
<!-- /issue-number -->

closes bpo-31608: Fix a crash in methods of a subclass of _collections.deque with a bad __new__(). (GH-3788)

The default value of asyncio.StreamReader *limit* is `_DEFAULT_LIMIT` instead of `None`.

<!-- issue-number: [bpo-34613](https://www.bugs.python.org/issue34613) -->
https://bugs.python.org/issue34613
<!-- /issue-number -->

<!-- issue-number: [bpo-28617](https://www.bugs.python.org/issue28617) -->
https://bugs.python.org/issue28617
<!-- /issue-number -->

Although the kernel accepts any negative value for timeout, the
documented value to block indefinitely is -1.

This commit also makes the code similar to select.poll.poll().

Rewritten/updated sections:

* Event Loop APIs
* Transports & Protocols
* Streams
* Exceptions
* Policies
* Queues
* Subprocesses
* Platforms

8621bb5d sets the filename in directly in the FileNotFoundError, so we may revert the earlier fix 5f780400.

Mention the implicit cache in struct.Struct() docs.

Consistent with the re.compile documentation note.

Emit warning when None passed explicitly, list Python version since
deprecation in warning message and docs.

Remove ellipsis that look like continuation prompts,
has a side benefit of putting rest of error message in proper text color.

This docstring has drifted since python2: https://github.com/python/cpython/blob/ca079a3ea30098aff3197c559a0e32d42dda6d84/Lib/encodings/__init__.py#L68

When subprocess.Popen() stdin= stdout= or stderr= handles are specified
and appear in pass_fds=, don't close the original fds after dup'ing them.

This implementation and unittest primarily came from @izbyshev (see the PR)

See also https://github.com/izbyshev/cpython/commit/b89b52f28490b69142d5c061604b3a3989cec66c

This also removes the old manual p2cread, c2pwrite, and errwrite closing logic
as inheritable flags and _close_open_fds takes care of that properly today without special treatment.

This code is within child_exec() where it is the only thread so there is no
race condition between the dup and _Py_set_inheritable_async_safe call.

Check for functionality via imports rather than checking sys.platform
specifically for Windows

* bpo-33217: Raise TypeError for non-Enum lookups in Enums

HMAC's digestmod was deprecated marked for removal, this removes it as planned.

Many type object initializations labeled a field "tp_size" in the
comment, but the name of that field is tp_basicsize.

The recursive frame pruning code always undercounted the number of elided frames
by one. That is, in the "[Previous line repeated N more times]" message, N would
always be one too few. Near the recursive pruning cutoff, one frame could be
silently dropped. That situation is demonstrated in the OP of the bug report.

The fix is to start the identical frame counter at 1.

also mention the change and its consequences in What's New

This reverts commit 10b59f1b.