Initialise them in the presumably intended way to get rid of the
warning. Note that the code around there is somewhat confused and
downright wrong anyway, though.

Also "+0", "-0" or "000" for instance are correct ways to write zero.
Also SAL_MAX_INT32 and SAL_MIN_INT32 are legal values for a UNO
"long". (And it will always be true that a sal_Int32 is <=
SAL_MAX_INT32 and >= SAL_MIN_INT32.) And does the code accept any
single-character string as zero?

Then there is copy-paste galore for the "hyper", "double" and "float"
cases.

Change-Id: Ie1920cfb4d17dca7cd966d8b200d2510ea975290