Anecdotal evidence suggests that a single EnumPrinters() call

Anecdotal evidence suggests that a single EnumPrinters() call
specifying both PRINTER_ENUM_LOCAL and PRINTER_ENUM_CONNECTIONS
catches more printers in some circumstances than two EnumPrinters()
calls each specifying just one of them. We'll try it for a bit; if
it goes wrong I might have to put back the two original calls as
well and sort out some means of removing duplicate printers from the
list.

[originally from svn r1829]