distributed application weird behaviour

[crownedgrouse]

Describe the bug
When a node y@b is listed in distributed option for a kernel application distribution and not belonging to at least sync_nodes_optional, main node x@a is waiting forever (infinity timeout ?) if y@b not started.

To Reproduce
Example of kernel configuration :

[{kernel,
  [
    {distributed, [{my_app, 5000, [x@a, y@b]}]},
    {sync_nodes_mandatory, []},
    %{sync_nodes_optional, [y@b]},
    {sync_nodes_timeout, 15000}
  ]}
 ].

Expected behavior
Maybe I missed this in documentation, and it is a (strange) feature, or it is a bug.
I would expect main node x@a to start anyway after timeout set, considering y@b node optional, or at least a failure for an invalid configuration.
Does it mean that all nodes listed in distributedmust be either in sync_nodes_mandatoryor sync_nodes_optional ?

note that below configuration is working as expected (x@a starting after timeout), while x@a not listed in sync_nodes_mandatoryor sync_nodes_optional :

[{kernel,
  [
    {distributed, [{my_app, 5000, [x@a, y@b]}]},
    {sync_nodes_mandatory, []},
    {sync_nodes_optional, [y@b]},
    {sync_nodes_timeout, 15000}
  ]}
 ].

Affected versions
24.0 (at least)