Add an icon (in OSX parlance, a 'bundle') to the installation folder to start
I2P.  

While there might be a better way to handle this (admittedly, I don't know OSX
that well), it is my belief that this way is less 'hackish' than the various
OSX 'installers' that I've seen floating around.

<os family=unix> covered OSX as well. 

Instead, we'll show the panel for everything *EXCEPT*
Windows.

(either wrong architecture or i2psvc is just plain missing) and give
a more useful error message.

Add another commented example to wrapper.config for those instances in which bindipv6only causes problems

Windows Uninstall the service if I2P is installed, whether the service was enabled by the installer or not.

My rationale for this:
1) in case someone is installing I2P on top of an already existing installation
   this will cause I2P to be shutdown
2) no hangs during the installation if i2p is already installed as a service
3) in case the NT service settings are changed in wrapper.config by us, this
   will force the service settings to be updated.
4) versions of I2P < 0.8.7 did not remove the service when uninstalling. If someone
   removed I2P in the past and reinstalls I2P--opting to run as a service--the
   installation process will hang.

    jbigi and wrapper files for arm.
    Compiled on trimslice with gcc version 4.4.5 (Ubuntu/Linaro 4.4.4-14ubuntu5) 
    Log postinstall errors to postinstall.log.
    java version "1.6.0_18"
    OpenJDK Runtime Environment (IcedTea6 1.8.7) (6b18-1.8.7-0ubuntu2.1)
    OpenJDK Zero VM (build 14.0-b16, mixed mode)
    GMP 4.3.2 (half the size of 5.0.2, and no speed difference) LGPLv3
    Wrapper 3.5.9 GPLv2
    All binaries stripped.

    As on the Android emulator, the stock BigInteger.modPow()
    has some serious bug on arm.
    Without the libjbigi checked in here, the JVM crashes almost immediately.

With this checkin I'm trying to lessen the occurences of ticket #474:

If a user installs I2P on top of an already existing I2P installation with the service
enabled, the installer will hang. The Quit button cannot be clicked. Clicking
the X in the corner seems to roll back the installation.

Also #472: service is not removed when I2P is uninstalled

  to do the first time that I noticed I added these files with those unneccesry
  attributes.
* Set the "manual merge" attribute on these files.

working directory set to $INSTALL_DIR, C:\Program Files\I2P is not removed in
XP.

...but even with the working directory removed, the the Uninstaller folder is left
within $INSTALL_DIR in Windows 7.

for use with a 32 bit JRE.

Rationale:
On an x64 system using a 32 bit jvm Without the 32 bit libwrapper, messages
like this will be shown in wrapper.log with the wrapper in MTN & I2P >= 0.8.7:

-----------------------
 Launching a JVM...
 WrapperManager: Initializing...
 WrapperManager: 
 WrapperManager: WARNING - Unable to load the Wrapper's native library 'libwrapper.so'.
 WrapperManager:           The file is located on the path at the following location but
 WrapperManager:           could not be loaded:
 WrapperManager:             $I2P/lib/libwrapper.so
 WrapperManager:           Please verify that the file is both readable and executable by the
 WrapperManager:           current user and that the file has not been corrupted in any way.
 WrapperManager:           One common cause of this problem is running a 32-bit version
 WrapperManager:           of the Wrapper with a 64-bit version of Java, or vica versa.
 WrapperManager:           This is a 32-bit JVM.
 WrapperManager:           Reported cause:
 WrapperManager:             $I2P/lib/libwrapper.so: $I2P/lib/libwrapper.so: 
                           wrong ELF class: ELFCLASS64 (Possible cause: architecture word width mismatch)
 WrapperManager:           System signals will not be handled correctly.
-----------------------

With libwrapper.so removed, one sees the following:

 WrapperManager: WARNING - Unable to load the Wrapper's native library because none of the
 WrapperManager:           following files:
 WrapperManager:             libwrapper-linux-x86-32.so
 WrapperManager:             libwrapper.so
 WrapperManager:           could be located on the following java.library.path:
 WrapperManager:             $I2P
 WrapperManager:             $I2P/lib
 WrapperManager:           Please see the documentation for the wrapper.java.library.path
 WrapperManager:           configuration property.
 WrapperManager:           System signals will not be handled correctly.
-----------------------

The 32 bit lib names, when installed on an x64 system, will match the alternate
names that the wrapper looks for.

Tested in Windows 7.
As has been customary, these were linked with gmp-5.0.2.

Built using 
"x86_64-w64-mingw32-gcc (GCC) 4.5.3" on Debian Sid with debug symbols stripped
using x86_64-w64-mingw32-strip.

The Tanuki Software website states "64-bit Windows versions of the Java Service Wrapper
are not currently being made available in the Community Edition." The Makefile
for x86_64 is missing from the upstream tarball as well. 

Well...included in this checkin is a diff against 
$WRAPPER-3.5.9-SRC/src/c/Makefile-windows-x86-32.nmake (see the README in
installer/libs/wrapper/win64.

- removing from /build.xml
- moving recent changes from installer/resources/postinstall.bat to installer/install.xml
- dropping installer/resources/postinstall.bat

    * JBigI:
      - Add new libjbigi-linux-xxx.so files built by sponge, now that we have
        CPUID.java support for them (thanks hottuna).
        See http://zzz.i2p/topics/306 for discussion and test results.

        64-bit processors (atom, core2, corei, nano):
        Built with GMP 5.0.2. License is LGPLv3.
        Built by sponge with GCC 4.4.4, downloaded from
        http://sponge.i2p/files/jbigi/gmp-5.0.2/
        For 64-bit processors, both performance testing and
        the GMP changelog http://gmplib.org/gmp5.0.html led us to use 5.0.2
        for both the 32- and 64-bit versions, even though the files are twice as big.
        5.0.x contains specific optimizations for atom and nano.
        All 64-bit libs have _64 appended.

        32-bit processors (pentiumm, geode):
        Built with GMP 4.3.2. License is LGPLv3.
        Built by sponge with GCC 4.4.4, downloaded from
        http://sponge.i2p/files/jbigi/gmp-4.3.2/
        The 4.3.2 files are half the size of those built with GMP 5.0.2, and there was
        little or no performance difference between the two versions for 32 bit processors.

Compiled using mingw32 in Windows XP (gcc --version returns v4.5.2).

As was done with the Linux and FreeBSD builds, jbigi was linked with gmp-4.3.2
for the 32bit cpus and 5.0.2 for the 64bit CPUs.

Since there current isn't a free version of the wrapper for x64 Windows, there
aren't any x64 versions of these dlls.

I missed in revision 97672ae2bf487e93c095bb1d142c3df1a2f3d7d7

*Sigh*

The aging v3.1.x of the java wrapper is being replaced in this
checkin with the latest (current) stable release, 3.5.9.

There is now only one copy of wrapper.jar, located in wrapper/all.
The build.xml scripts have been changed to reflect this new path.

All binaries--with the exception of solaris (I don't have 
SPARC hardware)--have been stripped.

FreeBSD x64 is being included in this release.

I have only included the x86 macosx wrapper due to not having a
system to test the x64 code on. On the plus side, the wrapper is 
universal (a FAT powerpc & Intel Mac) binary. No more prompting
to install Rosetta!

Drop libjbigi-freebsd-viac32.so, dup of libjbigi-freebsd-pentium3.so, case already handled in NBigI.

with PIC and stripped

Just as was done with Sponge's compiles for Linux, the 32bit libs/cpus were linked with gmp 4.3.2.

The 64bit compiles were linked with gmp 5.0.2. The 32bit compiles for 64bit cpus were linked with 5.0.2 as well.

All .so files were compiled with the default gcc on FreeBSD-8.2-RELEASE
gcc (GCC) 4.2.1 20070719  [FreeBSD]

      - Replace old non-PIC 62-bit linux libs (GMP 4.1.4 built in 2005)
        with PIC libs built with GMP 5.0.2. License is LGPLv3.
        Built by sponge with GCC 4.4.4, downloaded from
        http://sponge.i2p/files/jbigi/gmp-5.0.2/
        For 64-bit processors, both performance testing and
        the GMP changelog led us to use 5.0.2 for both the 32- and 64-bit versions,
        even though the files are twice as big.
        See http://zzz.i2p/topics/306 for discussion and test results.
        libjbigi-linux-athlon64.so was a 64 bit file; by the new naming standard
        supported by NativeBigInteger, it is now a 32-bit file and the
        64-bit one is libjbigi-linux-athlon64_64.so.
        All new 64-bit libs will have _64 appended.
      - Update build docs

        NBI now tries athlon as a fallback for all 64-bit processors.