Do not schedule bucket expiry again after initial response

In case multiple responses come in, it does not make sense to expire the
bucket again and again. We should only do this once for the initial
bucket, because otherwise on subsequent requests in the next ratelimit
window we might expire it, which is wrong.

Author: jchristgit

lib/nostrum/api/ratelimiter.ex

@@ -771,8 +771,6 @@ defmodule Nostrum.Api.Ratelimiter do
         %{outstanding: outstanding} = data
       )
       when remaining >= 0 do
-    expire_bucket = {{:timeout, bucket}, reset_after, :expired}
-
     case Map.fetch(outstanding, bucket) do
       # This is the first response we got for the absolute initial call.
       # Update the remaining value to the reported value.
@@ -781,7 +779,7 @@ defmodule Nostrum.Api.Ratelimiter do
 
         {:keep_state, %{data | outstanding: updated_outstanding},
          [
-           expire_bucket,
+           {{:timeout, bucket}, reset_after, :expired},
            {:next_event, :internal, {:next, remaining, bucket}}
          ]}
 
@@ -800,7 +798,6 @@ defmodule Nostrum.Api.Ratelimiter do
       {:ok, {stored_remaining, _queue}} ->
         {:keep_state_and_data,
          [
-           expire_bucket,
            {:next_event, :internal, {:next, stored_remaining, bucket}}
          ]}