10000 completes in a too short time and may easily provide unreliable
figures because of tiny duration.

bulk_data field size was not removed from the count. It is not possible
to declare it simply as 'char bulk_data[]' since the structure is nested
into another structure.

Improved getting pid

See https://sourceware.org/glibc/wiki/Release/2.20#Packaging_Changes

```sh
$ ~ pidof redis-server
# nothing

$ ~ ps aux | grep [r]edis
redis      593  0.0  0.0  36900  5564 ?        Ssl  Dec02   1:37 /usr/bin/redis-server 127.0.0.1:6379
klay     15927  0.0  0.0  16772  6068 pts/6    S+   13:58   0:00 redis-cli

$ ~ uname -a
Linux edge 3.17.4-1-ARCH #1 SMP PREEMPT Fri Nov 21 21:14:42 CET 2014 x86_64 GNU/Linux
```

This allows shell pipes to correctly end redis-cli.

Ref #2066

Thanks to @badboy for the help in checking the build after the fix.

Track bandwidth used by clients and replication (but diskless
replication is not tracked since the actual transfer happens in the
child process).

This includes a refactoring that makes tracking new instantaneous
metrics simpler.

Closes issue #1935.

Ref: issue #2175

PFCOUNT is technically speaking a write command, since the cached value
of the HLL is exposed in the data structure (design error, mea culpa), and
can be modified by PFCOUNT.

However if we flag PFCOUNT as "w", read only slaves can't execute the
command, which is a problem since there are environments where slaves
are used to scale PFCOUNT reads.

Nor it is possible to just prevent PFCOUNT to modify the data structure
in slaves, since without the cache we lose too much efficiency.

So while this commit allows slaves to create a temporary inconsistency
(the strings representing the HLLs in the master and slave can be
different in certain moments) it is actually harmless.

In the long run this should be probably fixed by turning the HLL into a
more opaque representation, for example by storing the cached value in
the part of the string which is not exposed (this should be possible
with SDS strings).

start_server now uses return value from Tcl exec to get the server pid,
however this introduces errors that depend from timing: a lot of the
testing code base assumed the server to be actually up and running when
server_start returns.

So the old code that waits to see the pid in the log file was restored.

It's hard to run the Redis test continuously if it leaks processes on
exceptions / errors.

There was no sane way to detect tests that may never end because of
Redis bugs or tests bugs.

Because of (not so) recent Redis changes, now the LRU internally
reported unit is milliseconds, not seconds, but the DEBUG OBJECT output
was still claiming seconds while providing milliseconds.
However OBJECT IDLETIME was working as expected, which is the correct
API to use.

This syncs lua-cmsgpack with the mattsta/lua-cmsgpack upstream.

Fixes #2161

zmalloc(0) cauesd to actually trigger a non-zero allocation since with
standard libc malloc we have our own zmalloc header for memory tracking,
but at the same time the returned pointer is at the end of the block and
not in the middle. This triggers a false positive when testing with
valgrind.

When the inline protocol args count is 0, we now avoid reallocating
c->argv, preventing the issue to happen.

Basically: test to make sure we can load cmsgpack
and do some sanity checks to make sure pack/unpack works
properly.  We also have a bonus test for circular encoding
and decoding because I was curious how it worked.

Main reasons for upgrade:
  - Remove a warning when building Redis
  - Add multi pack/unpack
  - Improve memory usage and use Lua allocator properly
  - Fix some edge case encoding/decoding bugs

clang doesn't like "extern inline" when no definition
is given right away.