dd: a zero factor no longer short-circuits the multiplier

[igorgbr]

GNU dd stops at the zero factor in a multiplier expression and never looks at
the rest. We now parse every factor first, so a huge one fails before the zero
is taken into account:

$ LC_ALL=C dd count=00x9999999999999999999999999999999999999999999999999999999999999 </dev/null
uutils: dd: invalid number: '00x999…': Value too large for defined data type
        exit=1
GNU:    0+0 records in
        0+0 records out
        0 bytes copied, 1.627e-05 s, 0.0 kB/s
        exit=0

This makes tests/dd/misc.sh fail.

It started with 7f9b9a6 ("dd: reject a number that does not fit in u64"),
which turned ParseSizeError::SizeTooBig into an error instead of u64::MAX
in parse_bytes_with_opt_multiplier. The zero factor case wasn't covered.

Note that fixing this alone doesn't make tests/dd/misc.sh pass the test then
reaches a second difference, dd: failed to seek in output file: Illegal seek.

Compared against GNU coreutils 9.10.