Relays now respect their AccountingMax bandwidth again. When relays
entered "soft" hibernation (which typically starts when we've hit
90% of our AccountingMax), we had stopped checking whether we should
enter hard hibernation. Soft hibernation refuses new connections and
new circuits, but the existing circuits can continue, meaning that
relays could have exceeded their configured AccountingMax.

This commit rolls back some of the cpu-saving fixes, where we tried
to avoid calling so many of our events while we're off the network.

That's because PERIODIC_EVENT_FLAG_NEED_NET checks net_is_disabled(),
which returns true even if we're only in soft hibernation.

Fixes bug 32108; bugfix on 0.4.0.1-alpha.

Since Rust on macOS is slow, don't wait for the macOS Rust job to finish.
Instead, split rust into slow rust (macOS) and fast rust (Linux). And
allow the build to finish before slow rust finishes.

Also make sure that we have:
* a Rust build on each platform,
* a Rust build with each compiler, and
* a check on all our Rust builds.

Finally, sort builds: allow fail last, macOS first, slowest first.

Closes 31859 for 0.3.5.

Part of 31859 for 0.3.5.

While merging:
* leave out some redundant jobs and build matrix entries

Since Travis macOS has IPv6 support (and Travis Linux does not), chutney
will now run its IPv6 networks as part of Travis CI.

But since chutney is slow, don't wait for the macOS chutney to finish.
(Travis have fixed the duplicate notification bug in fast_finish. So we
can use fast_finish and allow_failure to finish early. Unfortunately,
allow_failure also means we ignore failures in macOS chutney.)

Also make sure that we have:
* a compile on each platform, with each compiler,
* a check on each platform, and
* a check on each compiler.

Finally, sort builds: allow fail last, macOS first, slowest first.

Closes ticket 30860.
Closes ticket 31859 for 0.2.9.

Part of 31859.

The jobs should be the same, but Travis may display them differently.

Part of 31859.

"ours" merge to avoid picking up 0.2.9 version of ticket 31673.

tor_isinf() was new in 0.4.0, and also needs to look at the
same rules as clamp_double_to_i64()

Our code assumes that when we're configured to get IPv6 addresses
out of a TRANS_PF transparent proxy connection, we actually will.
But we didn't check that, and so FreeBSD started warning us about a
potential NULL pointer dereference.

Fixes part of bug 31687; bugfix on 0.2.3.4-alpha when this code was
added.

We used to do this on Windows only, but it appears to affect
multiple platforms when building with certain versions of GCC, and a
common pattern for defining the floating-point classifier functions.

Fixes part of 31687. I'm calling this a bugfux on 31687, when we
started suppressing these warnings on Windows.

Closes #31673

Closes #31673

When tor is missing descriptors for some primary entry guards, make the
log message less alarming. It's normal for descriptors to expire, as long
as tor fetches new ones soon after.

Fixes bug 31657; bugfix on 0.3.3.1-alpha.

tor-github/pr/1174, but with the last commit re-worded to remove
the fixup, because fixups break our push rules.

This change makes test-stem faster and more reliable.

Use "make test-stem-full" to run all of stem's tests.

Closes ticket 31554.

Sebastian Hahn authored Aug 20, 2019 and teor committed Aug 22, 2019

The main build doesn't use HARDENING_OPTIONS, so the exclude shouldn't,
either. Reduces the number of builders by one.