gb_SourceEnvAndRecurse_STAGE is unset in CustomTarget's make process
when the parent make has been started in sourced environment. This leads
to all sorts of "interesting" failures because the gbuild classes and
platform stuff are not loaded...

The breakage was due to the drawing layer re-work.  We do need to store
charts (or all OLE shapes with associated listening ranges) for later
registration regardless of anchor types.

Or else filtering of field members will not work correctly when the
field contains an empty name.

one test disabled because we first need to write a patch for our
internal cppunit

With this change, the automatic y-axis scaling should be more accurate
for stacked charts (i.e. stacked line, area, bar charts) with dates on
the x-axis.  The previous algorithm only calculated the max and min
y-values of the individual data, without taking into account the stacking
type when the x-axis type was date.

Or else we'd end up drawing a pretty interesting diagram...

(With the insight that Data::components is RootNode::members.)

Triggered by fdo#43725, incrementing an invalid iterator bombs -
though this seems a corner case, depends on whether one considers
"+= 0" as incrementing or not.

 * some cleanup - removed ugly-as-hell exposure of only parts of the
   internal data struct on B2DPolygon (and reworked the only client
   of that code)
 * added stl-style begin/end to B3DPolyPolygon as well, mirroring
   B2DPolyPolygon, plus adapting the insert() func to make use of
   that (avoiding loads of temporaries)

Though it looks wrong that rtl_allocateMemory(0) == 0.

Add --bzip2 and --xz options to select the compression. Both options
can be used togetger to generate both types of tarballs in one call.
XZ compession is used by default.

Also renamed "lo*build" strings to "lo*core" to fit the current name
of the top level source directory.