Fixes #1719

Fix #1663 by more gracefully handling non-type arguments.

The documentation specifically says that kill, killall, etc, all take a a type, but
allow passing exception instances.

Also allow passing arbitrary objects; this is also not documented, and not recommended.
The result will be raising a BaseException in the greenlet. That should be better than
silently printing something to stderr, which is what happened before.

Rework the way Semaphore deals with cross-thread usage

They seem stable now.

LDFLAGS is needed on the 2010 image (for clock_gettime). The 2014
image needs -pthread (for pthread_atfork, via cffi)

Also some tweaks to the make-manylinux script; we seem to not be getting the loop I expect.

Implement timeouts for the no-hub case using spin locks.

This should ultimately be far more robust, and it handles LoopExit cases much better.

Fixes #1698

Not yet complete; On Python 2, I sometimes get ValueError: over-released the semaphore for reasons I don't understand.

Also, Python 2 in general, and PyPy in particular, sometimes fails to make progress on the semaphore tests (one of the background threads doesn't complete for some reason). The RLock tests are much worse on Python 2 also. I don't understand why. Then again, I don't understand how the normal RLock could possibly actually be thread safe either: I think it must lean heavily on the GIL.

More gracefully handle errors pickling a traceback.

Fixes #1704.

Refs https://github.com/gevent/gevent/issues/1667

Python 2, subprocess: Let unbuffered binary writes to popen.stdin loop to write all the data

More like what Python 2 standard library does. Beware of relying on that though during an upgrade.

Fixes #1711

Surface the older changelogs more directly, and also surfate the module listing more directly.

Fixes #1700.

Bumping its 'noisy' value up, while there were plenty of missing-reference events it got called for, it never actually printed the line that showed it resolved something.

Avoid printing a TypeError traceback on certain types of invalid client requests

Specifically, those that contain a spurious % in the request line.

Test this, explicitly asserting both the returned status code and that we don't print a traceback.

Fixes #1708.

It's very slow, and we don't test it under manylinux anyway.

Remove references to Travis, and enable pypa/gh-action-pypi-publish to upload built wheels on tagged releases.

More tweaks. It took at least 33 minutes to build gevent on arm/python35. Enable some output to try to figure out why.

It doesn't exist, apparently.

Also, we got an error configuring c-ares back on the 2010 image for some reason. Enable its copious output to try to track it down.

Github Actions upgraded PyPy to 7.3.3 overnight so the tests broke.

TODO: Need to specify exact versions in the GHA config.