[lmi] Requesting Evgeniy's review

[Greg Chicares]

I'm trying to refactor all xml-library code, ideally so that
all calls to libxml++ occur in 'xml_lmi.?pp'. This may or
may not give us the option of substituting different xml
libraries, which may or may not be a good goal; but it is
educationally valuable to me, and I feel that the effort is
yielding clearer code. And safer code, too, because it's
easier to make sure all exceptions are trapped and all error
codes are checked. And above all it makes the transition
from xmlwrapp to libxml++ easier for me. Anyway, that's not
what I wanted to ask you about....

The last remaining use of this type:
    typedef std::list<xmlpp::Node*> NodeContainer;
(outside of 'xml_lmi.?pp') is in 'ledger_xml_io.cpp'. I made
tentative modifications there, in '#if 0' blocks so that I
wouldn't break anything. Could I ask you to review it over
the next day or two and let me know whether I've done it
correctly? It's different from other uses of 'NodeContainer'
because you are calling
    xmlpp::Node::get_children(argument)
with an actual argument, whereas elsewhere the argument is
just the default empty string. I've added such an argument
to this function
    xml_lmi::ElementContainer xml_lmi::child_elements
        (xml_lmi::Element const& parent
        ,std::string const& name
        )
but haven't studied 'double_formatter_t' well enough to be
sure I've done this correctly.