ffi: NULL char* does not reliably round-trip to empty string in TS

[cs01]

Bug

A null char* returned from FFI doesn't reliably compare === "" on the TS side. Worked around in lib/net.ts (PR #585) by exposing an explicit isOpen() probe instead of relying on empty-string-as-sentinel.

Impact

Bridges that return "error or no result" via a null C string can't use empty-string as a failure sentinel from the TS side. Forces each bridge to add explicit _is_* / _last_error probe functions.

Workaround pattern (current)

// net-bridge.c
char *cs_net_last_error(void *sock) { ... returns GC'd string or NULL ... }
int cs_socket_is_open(void *sock) { ... returns 0/1 explicitly ... }

// lib/net.ts — instead of
if (err === "") { /* ok */ }
// use
if (socket.isOpen()) { /* ok */ }

Likely fix location

FFI string-return marshaling — null char* should round-trip to TS as empty string or a distinguished null sentinel consistently.

[cs01]

Fixed by #593 (merged to main). FFI declare-function with char* return now coerces NULL to the interned empty string at the call-site, so bridgeReturnsNull() === "" is true. Test fixture tests/fixtures/edge-cases/ffi-null-string-to-empty.ts asserts.