The gated e2e suites run serial (-parallel 1), so every hardcoded time.Sleep adds directly to wall-clock. ~99s of literal second-sleeps across the gated tests (more in ms/const), and these fixed waits are also a flakiness source.

Fix: replace "sleep N then assert" with poll-until-condition (timeout N). Cuts serial wall-clock and de-flakes.

Files (gated: test/e2e/daemon/ready + test/sequentialtest):

• test/e2e/daemon/ready/banlist_e2e_test.go — 8× 5s @ lines 48, 83, 180, 286, 357, 453, 509, 557 (~40s, biggest single offender)
• test/e2e/daemon/ready/utxo_test.go:286 — 10s
• test/e2e/daemon/ready/multi_node_send_2_tx_test.go:58 — 10s
• test/e2e/daemon/ready/reorg_test.go — ~9s (incl :333)
• test/e2e/daemon/ready/smoke_test.go — ~8s (5 sleeps, incl :2097)
• test/e2e/daemon/ready/pruner_parent_before_child_test.go:204 — 5s
• test/sequentialtest/block_assembly/block_assembly_restart_test.go — ~6s (3 sleeps)
• test/e2e/daemon/ready/legacy_sync_test.go — ~4.8s (7 small)
• test/e2e/daemon/ready/block_persister_test.go — ~4s

Out of scope: test/e2e/daemon/wip/* sleeps (kafka_tls ~44s, unmined_tx_cleanup ~22s, invalid_block ~20s) — wip/ is not run by the smoke gate (smoke runs the ready/ package). Verify wip/ isn't run by a nightly before touching.

Related: #997 — the FSM-readiness ceiling in test/nodeHelpers/blockchainDaemon.go. Fixing it lets several of these sleeps be deleted outright rather than converted.