Add diagrams for stream coordinator

Author: acogoluegnes

deps/rabbitmq_stream/docs/stream_coordinator.md

@@ -0,0 +1,77 @@
+# Stream Coordinator
+
+## Single Active Consumer
+
+### "Simple" SAC (Not Super Stream)
+
+```mermaid
+sequenceDiagram
+    participant C as Coordinator
+    participant C1 as Connection 1
+    participant C2 as Connection 2
+    participant C3 as Connection 3
+    Note over C,C3: Simple SAC (not super stream)
+    C1->>C: register sub 1
+    C-)C1: {sac, sub 1, active = true}
+    activate C1
+    C1->>C1: consumer update to client
+    C2->>C: register sub 2
+    C3->>C: register sub 3
+    C1->>C: unregister sub 1
+    deactivate C1
+    C-)C2: {sac, sub 2, active = true}
+    activate C2
+    C1->>C1: consumer update to client
+    deactivate C2
+```
+
+### SAC with Super Stream Partition
+
+```mermaid
+sequenceDiagram
+    participant C as Coordinator
+    participant C1 as Connection 1
+    participant C2 as Connection 2
+    participant C3 as Connection 3
+    Note over C,C3: Super Stream SAC (partition = 1)
+    C1->>C: register sub 1
+    C-)C1: {sac, sub 1, active = true}
+    activate C1
+    C2->>C: register sub 2
+    C-)C1: {sac, sub 1, active = false, step down = true}
+    deactivate C1
+    C1->>C1: consumer update to client
+    C1->>C: active consumer in group
+    C-)C2: {sac, sub 1, active = true}
+    activate C2
+    C2->>C2: consumer update to client
+    C3->>C: register sub 3
+    Note over C, C3: active consumer stays the same (partition % consumers = 1 % 3 = 1)
+    deactivate C2
+```
+
+### `noconnection` management 
+
+```mermaid
+flowchart TB
+  A(monitor) --noconnection--> B(status = disconnected, set up timer)
+  B -. timeout .-> C(status = forgotten)
+  B -. nodeup .-> D(reissue monitors, send msg to connections)
+  D -. down .-> E(handle connection down)
+  D -. connection response .-> F(evaluate impacted groups)
+```
+
+* composite status for consumers: `{connected, active}`, `{disconnected,active}`, etc.
+* `disconnected` status can prevent rebalancing in a group, e.g. `{disconnected, active}` (it is impossible to tell the active consumer to step down)
+* consumers in `forgotten` status are ignored during rebalancing
+* it may be necessary to reconcile a group if a `{forgotten, active}` consumer comes back in a group ("evaluate impacted groups" box above).
+This is unlikely though.
+
+### Stale Node Detection
+
+```mermaid
+flowchart TB
+  A(RA) -- tick --> B(stale nodes = RA known nodes - cluster nodes)
+  B -. no stale nodes .-> C(nothing to do)
+  B -. stale nodes .-> D(remove connections from state)
+```