[Core] Refactor Alertmanager to retry unhandled alerts

We are currently creating a copy of each alert to avoid segfaults when
the next pop_alerts invalidates the lt alert objects we are holding
for handler callbacks.

However these alert copies are not deep enough since need to also
resolve the alert object methods e.g. message() therefore it is still
possible to result in lt segfaults.

The workaround is to check for any handlers not called, give them some
more time and eventually discard if still not handled.

Ref: https://github.com/arvidn/libtorrent/issues/6437

0 files changed, 0 insertions, 0 deletions