dash-spv: a header batch that fails validation is dropped and sync reports Synced at the stale tip #1194

HeadersPipeline::take_ready_to_store drains a segment's buffered headers before they are validated. When store_ready_batches / store_headers then rejects the batch (BlockHeaderValidator, or the "Segment chain break" check), sync_manager only logs the error and emits SyncEvent::ManagerError. After that:

• The rejected headers are gone, but the segment still counts as complete.
• Every later segment fails with "Segment chain break".
• Once all segments are drained, is_complete() is true, so finalize_sync_if_complete sets SyncState::Synced at the stale tip.
• The peer that served the batch is not penalised, and nothing is re-requested.
• A restart replays the same failure.

Reproduction. Run a real mainnet sync with #1161 and --start-height 750000. Checkpoint 750000 carries the wrong time and bits, so block 750024 is rejected. The client logs Headers sync complete at height 750000 and reports Synced 750000/2557756 while peers are at 2557756. With --start-height 0 it ends at Synced 32000/... the same way. #1161 only supplies the rejection. The pipeline and the error handling are unchanged from dev, so a batch that fails continuity or PoW on dev ends the same way.

Expected:

• A rejected batch is re-requested from another peer, and the peer that served it is penalised.
• Header sync never reports Synced below the best height its peers report.