#2 recode 3.6: garbage output

[Martin Trautmann]

Every 20th out of 20000 input lines was broken.

I did a simple conversion u8..l1 and reverted it via
l1..u8. The characters included German locations with
Umlaute such as äöüßÄÖÜ.

I did the test again on the diff of the first run.
Again, it was about one of 20 lines which caused
problems. Thus I guess it's not a problem of the
character tables, but of the memory allocation.

Sample data can be used e.g. from
http://prdownloads.sourceforge.net/opengeodb/opengeodb-0.2.3a-UTF8-text-orte.zip?download

>recode --version
Free recode 3.6
Written by Franc,ois Pinard <pinard@iro.umontreal.ca>.

SunOS 5.8

http://recode.progiciels-bpi.ca/showfile.html?name=dist/TODO
names
. + Memory leaks: see add_to_sequence, ALLOC,
recode_{m,re}alloc}.

Maybe that's the (unfixed) problem?