SIG needs to handle local BR being down

[kormat]

In the current code, if a local BR is unreachable, the SIG will timeout, but never receive a revocation. This implies that sciond will not get a revocation, so it will keep serving the same paths from its cache. If all paths for the SIG's destination happen to use the same BR, this situation will take up to 6 minutes to resolve (sciond caches paths for 5mins, pathmgr caches for another minute).

Until #1389 is done, the SIG needs to handle by telling sciond to flush its paths.

[scrye]

Gateway now only uses paths it detects as healthy, so a local router being down is no longer an issue if other healthy paths out of the AS exist.