Our enca code uses strdup() on the input encoding name, as we don't modify it we can use the original constant string. Uses less memory, code is simpler and faster. Fixes memory leak (noticed by ulion).