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

mlx/compile.cpp

@@ -352,7 +352,8 @@ void compile_simplify(
   // Helpers to identify identical scalars
   std::map<std::pair<uint64_t, Dtype::Val>, array> scalars;
   auto is_scalar = [](const array& a) {
-    return a.is_evaled() && a.ndim() == 0;
+    // Condition for when it's safe to read an array
+    return a.is_available() && a.ndim() == 0;
   };
   auto get_scalar_rep = [](const array& a) {
     uint64_t v = 0;