Increase accuracy of ParseMathMLUnsignedNumber Fraction result

The new test from d5e55d204b71710eb5eb5d2c683dd6698626df3c "tdf#144319:
sw_odfexport: Add unittest" started to cause UBSan failure during
CppunitTest_sw_odfexport (<https://ci.libreoffice.org/job/lo_ubsan/2136/>),

> /workdir/UnpackedTarball/boost/boost/rational.hpp:605:22: runtime error: signed integer overflow: 1073741824 * 2 cannot be represented in type 'int'
>     #0 0x2b450983594f in boost::rational<int>::operator/=(boost::rational<int> const&) /workdir/UnpackedTarball/boost/boost/rational.hpp:605:22
>     #1 0x2b450982e216 in Fraction::operator/=(Fraction const&) /tools/source/generic/fract.cxx:224:7
>     #2 0x2b45098312d3 in operator/(Fraction const&, Fraction const&) /tools/source/generic/fract.cxx:320:10
>     #3 0x2b46066963d5 in (anonymous namespace)::lcl_CountBlanks(MathMLAttributeLengthValue const&, int*, int*) /starmath/source/mathml/mathmlimport.cxx:1497:30
>     #4 0x2b46066943eb in (anonymous namespace)::SmXMLSpaceContext_Impl::startFastElement(int, com::sun::star::uno::Reference<com::sun::star::xml::sax::XFastAttributeList> const&) /starmath/source/mathml/mathmlimport.cxx:1526:25
[...]

Formula-14/content.xml in sw/qa/extras/odfexport/data/tdf144319.odt contains
width="0.444em", which resulted in the inaccurate Fraction 476741369/1073741824
since 67d83e40e2c4f3862c50e6abeabfc24a75119fc8 "speedup rational_FromDouble"
(which computes dVal=4.76741e+08 and nDen=1073741824; where before that speedup
it computed dVal=4.44e+08 and nDen=1000000000, which would have resulted in the
accurate Fraction 111/250).  The excessively large denominator (1073741824 =
std::numeric_limits<sal_Int32>::max()/2 + 1) then caused overflow later on, as
shown above.

Change-Id: I221549528e4fa155e10697d218ec76bf678e8b2f
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/122186
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>

0 files changed, 0 insertions, 0 deletions