Correct special winter handling in predictions

Correct special winter handling in predictions

The current serials prediction code has an exception for handling
cases where 'Winter 2010' is followed by 'Spring 2010' rather than
'Spring 2011'.  This exception assumed the more common case of a
combined subfield $y, e.g.:

$yps21,22,23,24

In doing so, it incorrectly handled less common cases where the $y
is listed separately:

$yps21$yps22$yps23$yps24

This change (with test data) ensures we only check the first $y
for our 'early winter' exception.

Signed-off-by: Dan Wells <dbw2@calvin.edu>
Signed-off-by: Mike Rylander <mrylander@gmail.com>