The assert in the bugreport is triggered by ScPrintFunc::CalcPages() passing
uninitialized values of nEndRow (and others). These variables apparently
get initialized only by constructors that take ScPrintState. These ctors
also set (the somewhat poorly named) bState and the call to CalcPages()
is guarded by this. However, GetPrintState() will simply create ScPrintState
filled with these uninitialized values and later on this will be used
with these ctors, so bState will be set, but nEndRow will be bogus.

Although 5217a2a0 introduced tdf#121439,
this strange bState logic and unitialized variables has been these since
the initial commit, and the code doesn't take any precautions to check
whether the values are valid or not, so I assume this always was just lucky
enough to work and 5217a2a0 finally triggered a problem.

Given that it's rather unclear to me how this is supposed to work properly,
just add an extra flag to both ScPrintFunc and ScPrintState marking whether
the values are set or not and make CalcPages() depends on this flag instead.

Change-Id: I0620de6562865c24f5a0edca2566b01546bf2e2b
Reviewed-on: https://gerrit.libreoffice.org/68739Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
Tested-by: Jenkins
(cherry picked from commit 9432bab9)
Reviewed-on: https://gerrit.libreoffice.org/69261Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>