Stale LightClientFinalityUpdates sent during sync

[michaelsproul]

Describe the bug

It seems that Lodestar propagates stale LightClientFinalityUpdates on gossip. I suspect these are from syncing nodes.

I added some logging to Lighthouse to catch the updates we are rejecting, and I've found several from Lodestar that were very old:

• received finality update from slot 3567251, but expected 3660662
  that's 93411 slots behind
  sending peer: Lodestar v1.25.0, 16Uiu2HAkwepnLHcFXrWzvMPAmXrQSpMdoQg42Z4aguHx51kiftht
• received finality update from slot 3567136, but expected 3660619
  that's 93483 slots behind
  sending peer: Lodestar v1.25.0, 16Uiu2HAkxCFSKMbWnShN75PJvgmyZb6hpJJYqjNU1czStjd7yZwW

Expected behavior

Lodestar shouldn't publish stale light client updates if it knows it's syncing. Perhaps there's an argument to say this behaviour is harmless, in case the node is eclipsed and has been tricked into thinking it is syncing when it isn't? However, given that these are finality updates, they are likely only useful if the node is not eclipsed (i.e. is part of the 2/3+ of validators that are finalizing the chain).

Steps to reproduce

No response

Additional context

No response

Operating system

Linux

Lodestar version or commit hash

v1.25.0

[michaelsproul]

This was on Holesky btw