pico w: setsockopt: Fix sense of disable vs enable

memcmp() returns -1, 0 or 1 to denote the relative ordering of the two
buffers. So, the computation would actually set `enable = 0`
in the case where `value` had the same bits set as "one" and `enable = 1`
in the case where `value` had any other bits.

By changing the compared buffer to be `zero`, `enable` gets a true value
whenever the value is NOT exactly 0 (e.g., it's 1, 7, -1, ...),
correcting the sense of enable vs disable.

Thanks to @anecdata for testing and finding this problem, which previously
would have affected the nodelay flag as well.

Author: jepler

ports/raspberrypi/common-hal/socketpool/Socket.c

@@ -1184,8 +1184,8 @@ void common_hal_socketpool_socket_settimeout(socketpool_socket_obj_t *self, uint
 }
 
 int common_hal_socketpool_socket_setsockopt(socketpool_socket_obj_t *self, int level, int optname, const void *value, size_t optlen) {
-    int one = 1;
-    bool enable = optlen == sizeof(&one) && memcmp(value, &one, optlen);
+    int zero = 0;
+    bool enable = optlen == sizeof(&zero) && memcmp(value, &zero, optlen);
 
     switch (level) {
         case SOCKETPOOL_IPPROTO_TCP: