This affects the strength of the "embolden" flag, since the default
is too heavy.

Can be used in a fonts.conf:

<match target="font">
<test name="family"><string>Input Mono</string></test>
<edit name="embolden" mode="assign"><bool>true</bool></edit>
<edit name="embstrength" mode="assign"><int>20</int></edit>
</match>

Or in an xft line:

xterm -fa 'input mono:embolden:embstrength=10'

This change looks more invasive than it is because of having to
re-generate fcobjshash.h from gperf (which looks like it was out of
sync with fcobjshash.gperf anyway?)

This reverts commit 287a26070cc7cb08d9bdce76009fe4cd0af1bb62.

So that your statistics remain correct if the system wall clock is
changed during playback.

Use CLOCK_UPTIME instead of CLOCK_MONOTONIC so that statistics remain
correct across suspend/resume.

Testing by tobias@, sthen@,
and Benjamin Baier <programmer AT netzbasis DOT de>.

ok sthen@

Bulk ports build checked by naddy@. Thanks.

* The Xorg process could crash when multiple primary screens are
configured in xorg.conf.
* TearFree could trigger debugging messages in the pixman library

ok jsg@

Avoids failing page flips reported by landry@ with a pending xfce4 4.13
update (doesn't occur with xfce 4.12 currently in ports).

clock_nanosleep() isn't available yet so the usleep() path stays for
os_time_sleep()

Avoids diffs in inteldrm, libdrm and Mesa >= 17.2.

ok kettenis@

ok okan@

It was needed at some point in the past, but doesn't compile and
isn't needed with clang. Reported by jsg@.

with this change there is no need to manually do any randr commands
if your desktop environment supports it (gnome, mate, kde, etc.)

ok matthieu@, kettenis@

Odd behaviour reported by Ve Telko.