bug: race condition due to multiple backend connections being open

[0x416e746f6e]

at the periods of high loads rollup-boost tends to open multiple (dozens) of TCP connections to the builder, which can deliver otherwise sequential updates from sequencer in arbitrary order.

one example timeline (from rollup-boost logs):

2025-03-28T13:31:15.296Z,"{""level"":""INFO"",""fields"":{""has_attributes"":true,""head_block_hash"":""0xf7b308e151bf7c13aa362494ab499d3eee3c634a5eb0a1228e9e1cd44e874c69"",""message"":""received fork_choice_updated_v3""},""timestamp"":""2025-03-28T13:31:15.296585Z"",""target"":""rollup_boost::server""}"
2025-03-28T13:31:15.297Z,"{""level"":""INFO"",""fields"":{""method"":""\""engine_getPayloadV3\"""",""message"":""proxying request to rollup-boost server""},""timestamp"":""2025-03-28T13:31:15.297978Z"",""target"":""proxy::call""}"
2025-03-28T13:31:15.298Z,"{""level"":""INFO"",""fields"":{""payload_id"":""0x03e01ec8a970741a"",""message"":""received get_payload_v3""},""timestamp"":""2025-03-28T13:31:15.298012Z"",""target"":""rollup_boost::server""}"
2025-03-28T13:31:15.318Z,"{""level"":""ERROR"",""fields"":{""local_payload_id"":""0x03e01ec8a970741a"",""external_payload_id"":""0x03e01ec8a970741a"",""message"":""error calling get_payload_v3 from builder"",""error"":""ErrorObject { code: ServerError(-38001), message: \""Unknown payload\"", data: None }"",""url"":""http://34.162.194.218:8651/""},""timestamp"":""2025-03-28T13:31:15.318342Z"",""target"":""rollup_boost::server""}"
2025-03-28T13:31:15.318Z,"{""level"":""INFO"",""fields"":{""number"":""3098535"",""payload_id"":""0x03e01ec8a970741a"",""context"":""l2"",""message"":""returning block"",""hash"":""0x9bb4f13f944d146aac771d7304230f3a07b5bf9c50ab3da78e88fce046de360f""},""timestamp"":""2025-03-28T13:31:15.318365Z"",""target"":""rollup_boost::server""}"

related logs from the builder side (timestamps are for when the response was sent back):

related to: #126

[avalonche]

there's only a 10ms difference between the fcu and getPayload request. I think by the time the fcu response was completed another getPayload request was received, initiating a new request. Is this not a bug caused by op-node having too little time between requests?

[ferranbt]

Closing this since #206 fixed the "Unknown Payload" issue.