i.e StgDataStrm::Read takes the bool of no error and
multiplies it by the block size to report the length
read. A short read isn't an error so full buffer is
considered valid.

To keep #i73846# working and get deterministic fuzzing
results, zero out the trailing space of a successful but
short read. Changing this to return the truthful
number of bytes read makes #i73846# stop working.

There's wonderful nonsense here calculating nPg2,
determining nBytes to read derived from this,
reading nBytes into the buffer and then considering
it an error if nPg2 is not 1 after the presumably
potentially fatal buffer overflow if nPg2 wasn't initially
1, but this doesn't seem possible in practice

Change-Id: I2bac6066deb8a2902677e04696367ba2c351b325
Reviewed-on: https://gerrit.libreoffice.org/44310Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>