issue #93: cleaning up memory mapped files left over in cases where FILE_FLAG_DELETE_ON_CLOSE will not clean up file

issue #112: Adding permission for  NETWORK_SERVICE on the folder redis-server.exe is in during --service-install.  This allows for RDB files to be written and for .conf files to open in RW mode. Fixed a documentation typo.

Conflicts:
	bin/release/redis-2.8.4.zip
	src/Win32_Interop/Win32_QFork.cpp
	src/sds.c

Conflicts:
	00-RELEASENOTES
	deps/linenoise/linenoise.c
	src/anet.c
	src/debug.c
	src/redis-cli.c
	src/redis.c
	src/sentinel.c
	src/t_zset.c

Issue 108: Added 32-bit support back in. QFork code requires a contiguous memory allocation for heap. Due to this requirement and 32-bit memory model limitations, maxheap is limited to about 512MB in 32-bit executable. 32-bit version appears to be about 30% slower than 64-bit version on 64-bit OS.

issue 99: adding sample to demonstrate config of master + 3 slaves + 3 sentinels on one box. Discovered along the way that maxheap would not propagate through .conf include statements. It does now.

fix for issue #97: Added a modified sample to demonstrate large async GET operation. Also discovered that reply reading in async.c would stop if the reply data exceeded the buffer allocated for the read. redisAeAddRead() would not reschedule a read if already in read mode. To fix this is added redisAeForceAddRead().

Issue #95: Adding ability to specify custom service name/have multiple Redis service instances running.(Forgive the whitespace changes in Win32_Service.cpp. My editor was set to keep tabs when I wrote the code originally. Tabs are gone now.)

Added support for SIO_LOOPBACK_FAST_PATH (http://msdn.microsoft.com/en-us/library/windows/desktop/jj841212%28v=vs.85%29.aspx) . This speeds up localhost performance on Win8/Server2012+ significantly. In order to take advantage of this clients will have to enable socket fast loopback before connect().

The internal HLL raw encoding used by PFCOUNT when merging multiple keys
is aligned to 8 bits (1 byte per register) so we can exploit this to
improve performances by processing multiple bytes per iteration.

In benchmarks the new code was several times faster with HLLs with many
registers set to zero, while no slowdown was observed with populated
HLLs.

When the register is set to zero, we need to add 2^-0 to E, which is 1,
but it is faster to just add 'ez' at the end, which is the number of
registers set to zero, a value we need to compute anyway.

adding working hiredis async example to solution based off of example-ae.c source. For issues 18 + 86

fdapi not properly mapping fstat in config.c, so we ended up invoking CRT version without the rfd->fd translation resulting in the CRT calling exit(). Fixes issue 43.

Given that the code was written with a 2 years pause... something
strange happened in the middle. So there was no function to free a
lex range min/max objects, and in some places the range was passed by
value.

Like ZCOUNT for lexicographical ranges.

Converts HyperLogLogs from sparse to dense. Used for testing.

After running a few benchmarks, 3000 looks like a reasonable value to
keep HLLs with a few thousand elements small while the CPU cost is
still not huge.

This covers all the cases where the dense representation would use N
orders of magnitude more space, like in the case of many HLLs with
carinality of a few tens or hundreds.

It is not impossible that in the future this gets user configurable,
however it is easy to pick an unreasoable value just looking at savings
in the space dimension without checking what happens in the time
dimension.