IPFS latency

when new nodes (client) added to our network, IPFS connections still can prevent validator from collecting or processing txs. We should just keep ipfs hash in lostChunk and update pending
since now ipfs in tx no longer contains critical infomation.

Start by tracing the validator’s transaction collection and processing flow, then locate the handling of lostChunk, pending, and IPFS hashes in transactions. Done means IPFS latency from newly added client nodes no longer prevents transaction collection or processing, while the hash is retained in lostChunk and pending is updated.

Written by the indexing model from the issue text.