Right now, it is used to detect blocked event loops, and it is extensible by users. In the future there will be some more default monitoring options (e.g., memory).

Refs #1021.

Make more performance-sensitive places use _get_hub_noargs, since under Python 3 it is twice as fast to call as get_hub.

Introduce GEVENT_TRACK_GREENLET_TREE to disable greenlet tree features

As a performance optimization for applications where spawning
greenlets is critical. Plus some other optimizations to speed up
spawning in the general case.

CPython 3.6 with 1.2.2 vs these changes with tracking disabled:

| Benchmark              | 36_122_bench_spawn | 36config_bench_spawn_tree_off |
+------------------------+--------------------+-------------------------------+
| eventlet spawn         | 12.6 us            | 12.2 us: 1.04x faster (-4%)   |
| eventlet sleep         | 5.22 us            | 4.97 us: 1.05x faster (-5%)   |
| gevent spawn           | 4.27 us            | 5.06 us: 1.19x slower (+19%)  |
| gevent sleep           | 2.63 us            | 1.25 us: 2.11x faster (-53%)  |
| geventpool spawn       | 9.00 us            | 8.31 us: 1.08x faster (-8%)   |
| geventpool sleep       | 4.82 us            | 2.83 us: 1.70x faster (-41%)  |
| geventraw spawn        | 2.51 us            | 2.81 us: 1.12x slower (+12%)  |
| geventraw sleep        | 649 ns             | 679 ns: 1.05x slower (+5%)    |
| geventpool join        | 3.47 us            | 1.42 us: 2.44x faster (-59%)  |
| geventpool spawn kwarg | 11.0 us            | 8.95 us: 1.23x faster (-19%)  |
| geventraw spawn kwarg  | 3.87 us            | 4.20 us: 1.08x slower (+8%)   |

The differences compared to master are hard to quantify because the
standard deviation ends up being more than 10% of the mean in many
cases---and about a 10% improvement is what we typically see, so it
goes back and forth.

Also monkey-patch threading._Event on Python 2. Fixes #1136

Use strongly typed watcher callback functions in libuv, preventing compiler warnings

Turns out we weren't passing the macros we thought we were at all. Whoops.

And also ensuring that in the Python code we know the type of pointer we're dealing with. This might catch some errors.

Fix documentation for Pool.full()

The original docs had the purpose of the boolean return value reversed. I fixed that distinction, and also tried to clarify how the method can be used.

Exclude 3.3, account for cython being a develop wheel right now.

[skip ci]

Run tests in 3.7 dev mode; cleanup resource warnings

Especially as reported on PyPy.

The testrunner will now show the output of a successful test if the
output contains ResourceWarning.

Fix a bunch of such warnings.

Import _testcapi in testrunner to try to avoid the concurrency issues on PyPy.

Update to 3.7.0b2

This matches what all versions of Python do and reverts
https://github.com/gevent/gevent/commit/9b1bccffc11455112076189f35023291cf97a2a2

The bug that commit references was marked as "won't fix". This is part
of an effort to adapt to changes in 3.7b2 that lead us to produce a
different errno than the system does for
`socket.create_connection(('localhost', ...))`

Interpret -1 to select.poll.poll as blocking, especially under libuv

Why are we bothering to run cares on PyPy? It has several known-bad tests, and it sometimes crashes the interpreter. We don't recommend or support it now that we have dnspython.