MM-58038: Use context to call master for DeletePost (#27098)

Calling app.DeletePost immediately after creating a post
is susceptible to replica lag because we were calling the
replica to check for the post.

We fix this by passing a context to always query master.

https://mattermost.atlassian.net/browse/MM-58038

```release-note
NONE
```

server/platform/services/sharedchannel/sync_send.go

@@ -322,7 +322,7 @@ func (scs *Service) handlePostError(postId string, task syncTask, rc *model.Remo
 	}
 
 	// this post failed as part of a group of posts. Retry as an individual post.
-	post, err := scs.server.GetStore().Post().GetSingle(postId, true)
+	post, err := scs.server.GetStore().Post().GetSingle(request.EmptyContext(scs.server.Log()), postId, true)
 	if err != nil {
 		scs.server.Log().Log(mlog.LvlSharedChannelServiceError, "error fetching post for sync retry",
 			mlog.String("remote", rc.DisplayName),