Reverting regression with f0bf2a2a

configurable at build time via ABOUTBOXPRODUCTVERSIONSUFFIX in
instsetoo_native/util/openoffice.lst

configurable at runtime via registry/data/org/openoffice/Setup.xcu
and ooSetupVersionAboutBoxSuffix property

post salhelper visibility markup and gbuildification express
2008 doesn't link without salhelper mentioned on link line

the git-new-workdir helper is mostly useful to have one pristine clone
where the updates get pulled into & from that you create working copies.
As pulling to that master copy only updates the current branch (usually
master), checking out a different branch using git-new-workdir will
result in "your branch is behind the remote branch by xx commits" when
there have been commits after the branch was first checked out locally.

Thus this change adds a fast-forwarding merge when a branch is given.

so it does nothing on platforms where soversions are not used

renaming previous test to XRangeNames
new test for XRangeName

Turns out our encrypted zip package code writes out mismatching
compression methods in central directory and local file headers -
so not only be lenient towards the apple bug (n#653688), but also
ours (i.e. keep not comparing nMethod with nHow).

While fixing n#653688, turned out we write method=STORED into local
file headers, and method=DEFLATED into central directory structure
for encrypted ODF. Damn side-effects - fixed as hopefully intended
here.

Clearly this iteration and reading of the referer, that is then
subsequently not used at all by the method, can go.

This fixes cross-compiling from Linux to Android. My guess is that the
build system nowadays, when building the cross_toolset (for the build
platform), sources Env.Build.sh in an environment where Env.Host.sh
has already been sourced. So, if Env.Host.sh exports SYSBASE, that
stayed in force also after sourcing Env.Build.sh, with predictably
horrible results.

For some reason it didn't hurt cross-compiling from Mac OS X, which is
why I hadn't noticed.