Validator without ACKs ?

[SurfingNerd]

We found evidence in cases where validators can join a validator set, without writing acks.

example:
epoch switch:   3035
Block:  93504
Validator: 0x396Aa74164dc9bfF256f994C725155DD456fCC43
Did send his ACKs but those got never included in a block.
 http://explorer.uniq.diamonds/tx/0xc56cf78664d550476ce631a1d4c22e016a1e9b8a50e4e6835207c9ad7f98c2a9
 the transaction got replaced (nonce rule) later by this transaction:
 http://explorer.uniq.diamonds/tx/0xb5d8c525a2e9c398bb775a79e86a13b8acae137c157b509c196f534b6b495099/internal-transactions

Is it possible that those ACKS are still required for creating valid HBBFT messages ?
Maybe connected to #64 ?

[SurfingNerd]

in progress: Debug printing the count of parts and acks in SurfingNerd@7242e2f

[SurfingNerd]

observed another case:
epoch switch at 104450
ACK tx 0x3e4f45027c924f9b4a803162cee14f205a35ea4fc460a5fa3a28884f6f1b577f from 0xF001FaaA1aee5Cc972838766304398B97e909B63 could not get included.
despite that, the node got added to the validator set, but was kicked on epoch later.

http://explorer.uniq.diamonds/blocks/104450/transactions

[SurfingNerd]

Fixed with PR73
Released with v3.2.5-hbbft-0.8.0