Avoid intermediate computation overflowing sal_uInt16

At least during CppunitTest_sw_uwriter, some m_nStrResId is read so that Clang's
-fsanitize=implicit-signed-integer-truncation warned that an "implicit
conversion from type 'int' of value 74752 (32-bit, signed) to type 'sal_uInt16'
(aka 'unsigned short') changed the value to 9216 (16-bit, unsigned)".

Change-Id: I39f12d931bb18eea909494c8d7c9c0348fb71f27
Reviewed-on: https://gerrit.libreoffice.org/68664
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>

sw/source/core/doc/tblafmt.cxx

@@ -981,11 +981,9 @@ bool SwTableAutoFormat::Load( SvStream& rStream, const SwAfVersions& rVersions )
         {
             rStream.ReadUInt16( m_nStrResId );
             // start from 3d because default is added via constructor
-            sal_uInt16 nId = RES_POOLTABLESTYLE_3D + m_nStrResId;
-            if( RES_POOLTABLESTYLE_3D <= nId &&
-                nId < RES_POOLTABSTYLE_END )
+            if( m_nStrResId < RES_POOLTABSTYLE_END - RES_POOLTABLESTYLE_3D )
             {
-                m_aName = SwStyleNameMapper::GetUIName(nId, m_aName);
+                m_aName = SwStyleNameMapper::GetUIName(RES_POOLTABLESTYLE_3D + m_nStrResId, m_aName);
             }
             else
                 m_nStrResId = USHRT_MAX;