fix(e2e): make attested invalid proposal slot selection robust

[AztecBot]

Summary

Fixes the merge-queue failure that dequeued aztec-packages#24331. The failed merge-group run was not a merge conflict; x5-full failed in yarn-project/end-to-end/src/e2e_slashing/attested_invalid_proposal.test.ts because advanceToEpochBeforePipelinedTargetSlot only accepted the second slot of the next epoch. In the failing run, the bad proposer was repeatedly assigned to the first slot instead.

This changes the helper to scan all slots in the next epoch, choose the first slot assigned to the target proposer, and warp to the pipelined build slot (targetSlot - PROPOSER_PIPELINING_SLOT_OFFSET) before waiting for the bad proposal.

Verification

• ./bootstrap.sh ci from repo root: fails immediately because this branch has no root ci command (Unknown command: ci).
• yarn-project/bootstrap.sh format --check end-to-end: pass.
• yarn-project/bootstrap.sh compile: pass.
• Focused e2e attempted with yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_slashing/attested_invalid_proposal.test.ts "slashes a lazy attester for an invalid checkpoint and clears it on delayed equivocation", but local execution is blocked by missing generated protocol/L1 artifacts (protocol_contract_data.js, @aztec/l1-artifacts, Noir contract JSON). The dependency-aware root build needed to produce them is blocked in this container by Noir's just@1.42.4 install failing to reach crates.io through the proxy.

---

Created by claudebox · group: slackbot

[spalladino]

Closing in favor of #24339, which keeps the existing epoch-boundary selection logic and only increases the attempt window.