Reachability refactor

The result of Oban jobs determine the reachability status.

Publisher jobs will cancel themselves at execution time if the target server is now unreachable.

Receiving activities does not immediately mark a server as reachable, but creates a ReachabilityWorker job to validate.

A Cron will execute daily to test all unreachable servers.

test/pleroma/object/fetcher_test.exs

@@ -6,7 +6,6 @@ defmodule Pleroma.Object.FetcherTest do
   use Pleroma.DataCase
 
   alias Pleroma.Activity
-  alias Pleroma.Instances
   alias Pleroma.Object
   alias Pleroma.Object.Fetcher
   alias Pleroma.Web.ActivityPub.ObjectValidator
@@ -250,17 +249,6 @@ defmodule Pleroma.Object.FetcherTest do
       result = Fetcher.fetch_object_from_id("https://example.com/objects/no-content-type")
       assert {:fetch, {:error, nil}} = result
     end
-
-    test "it resets instance reachability on successful fetch" do
-      id = "http://mastodon.example.org/@admin/99541947525187367"
-      Instances.set_consistently_unreachable(id)
-      refute Instances.reachable?(id)
-
-      {:ok, _object} =
-        Fetcher.fetch_object_from_id("http://mastodon.example.org/@admin/99541947525187367")
-
-      assert Instances.reachable?(id)
-    end
   end
 
   describe "implementation quirks" do