Track the null leader_id reported during leader switch

[xiaoxichen]

During leader switch SM can report null leader_id to CM, it should all the way down to HS even to NuRaft Messenger.

[xiaoxichen]

If leader hb expires on follower, the follower will start pre-vote and set the leader to -1, which will then resulting SM report a "" (default value) in leader_id. It is very likely , the leader is still there, just the follower had a hiccup. The pre-vote will be rejected and other SM, might report correct leader_id to SM shortly.

This seems like a design more like a bug, from this follower's point of view, it believe the leader has gone and it started the leader election, there is nothing wrong for it to respond a "unknown " or "".

However, this create unnecessary PG update propagation in CM. We should ignore those leader_id = "" report in CM.

[xiaoxichen]

cc @jiankyu

[xiaoxichen]

example

[01/17/26 08:05:57.257330-07:00] [info] [93] [pg.cpp:284:on_process_hb_pg_status] pg.520 actual leader changed from 'f5810c97-c78e-4f54-9b80-05e134a31379' to '', reported by sm 2b0bb3b4-8a85-4261-b06c-ad0b45a370ee  

[01/17/26 08:06:24.879890-07:00] [info] [93] [pg.cpp:284:on_process_hb_pg_status] pg.520 actual leader changed from '' to 'f5810c97-c78e-4f54-9b80-05e134a31379', reported by sm 190cae7b-bc0f-4dd6-b94e-994e097787f4