Fixed incorrect behavior of Tracking4D with exclude_from_seed
Consider an event with clusters in two detectors, one being excluded from track seed. Then corry would try to build a seed track of two identical clusters (from the non-excluded detector), causing a crash.
Replaced a check for that, now instead of tree.size() < 2
it is reference_first == reference_last
(both null if there's no hits or equal if there's just one eligible hit).
Solves #181 (closed)