test(e2e): event handler can re-enter the library with a new request

[Ivansete-status]

What

Adds a C++ end-to-end test for re-entrancy: from inside an on_echo_fired event handler, the consumer issues another request to the library (carrying information taken from the event) and receives a correct response — no deadlock, no UAF.

Why this shape

The event handler runs on the FFI thread with the event-registry lock held (dispatchFFIEvent in ffi_events.nim). The test pins down the safe contract:

• ✅ async request from the handler — only queues work on the FFI request channel and returns immediately; the FFI thread drains it once the handler returns. The future is moved out of the handler and resolved on the main thread.
• ❌ a synchronous request would self-deadlock (the FFI thread is busy running the handler).
• ❌ add/removeEventListener from the handler would deadlock on the registry lock.

A one-shot guard avoids the echo → on_echo_fired → echo storm (echo re-fires the event). The wait_for timeouts turn any future deadlock regression into a test failure instead of a hang.

Test

Full C++ e2e suite 20/20 (the new EventHandlerCanIssueAsyncRequest plus the existing 19).

Note: while validating I hit a flaky-looking CrossLibrary SIGBUS locally — root-caused to a stale ./libecho.dylib that the incremental e2e build didn't rebuild (a clean checkout rebuilds it and the suite is green). Pre-existing build-hygiene nit, unrelated to this change; left out of scope.

🤖 Generated with Claude Code