I think there are two noexcept issues with control flow.

1. NAPI_THROW noexcept control flow

Like NAPI_THROW_IF_FAILED*, NAPI_THROW should also always abort control flow, but noexcept definition doesn't have a return so execution continues. All implementations using NAPI_THROW have a trailing return statement, so there is no bug, but I'd say it's unclean.

// for noexcept
#define NAPI_THROW(e)  (e).ThrowAsJavaScriptException();

2. noexcept ThrowAsJavaScriptException may produce infinite recursion

if napi_throw fails, it's likely that all further calls are failing and every time invoke another ThrowAsJavaScriptException, so I think this is a fatal case. Recursion is not a good idea, and also ignoring it, so it should fatal abort.