Investigate how we can automatically mark implicit withdrawals on hijacks with corrected paths about artemis HOT 2 CLOSED

NOTE: This may require a separate ARTEMIS sub-project to fully address; the goal of this issue is to investigate what it will take to implement, and if easy enough, implement it.

from artemis.

I propose the following:

1. detector sends benign updates also to a new taps (or detection function) called "implicit_withdrawal_generator"
2. implicit_withdrawal_generator checks the BGP update against the redis DB
3. if the (prefix, monitor_peer) pair of the update belong to an ongoing BGP hijack, it generates a fake withdrawal corresponding to the benign update and sends it as a normal monitor message to be processed by the db and the detector

We also need support in redis for marking all (prefix, peer) pairs that correspond to ongoing hijack events (and remove them when the hijack ends).

It probably needs some more thought, but the main concept is that with this mechanism we will be able to finally mark automatically the end time of a hijack (even if the hijacker does not explicitly withdraw a prefix, but the polluted paths towards the hijacker are simply cleaned up by changing routes to the legal/benign destination).

from artemis.