Shared events for synchronization + async eval (#998)

* more async eval

* fix rebase

* try correct async eval

* fix async

* more tests for async eval

* use shared events for synchronization

* comment + cleanup

* with autorelease pool

* fix no metal build

* fix compile

* fix patch

* don't eval if asyn evale'd

* don't use is_evaled

* comments

* more multi stream tests

* try and cleanup use of is_evaled

* use a status flag

python/src/convert.cpp

@@ -104,8 +104,7 @@ template <typename Lib, typename T>
 nb::ndarray<Lib> mlx_to_nd_array(
     array a,
     std::optional<nb::dlpack::dtype> t = {}) {
-  // Eval if not already evaled
-  if (!a.is_evaled()) {
+  {
     nb::gil_scoped_release nogil;
     a.eval();
   }