feat: if `SDXLC_URL` isn't reachable during a topology update it won't try to recover from it

[viniarck]

Currently, when handling a topology update, if this POST fails, the error handling is only logging as an error, which is a good start, BUT, depending how critical is it is to still have an outdated topology, you might consider the following:

1. Let this on_topology_updated_event handler crash, and then leverage kytos-ng core dead queue endpoint PATCH dead_letter/ try to reinject crashed events, hoping that eventually it succeeds, and stop trying to do it if a newer update happend or it succeeded.

2. Try to have a sort of reconciliation mechanism where it would try to poll the currently topology in a low frequency if any error happened, and only do it again if a POST to sdx topo error happens.