test(cheat-codes): retry warpL2TimeAtLeastTo in-current-slot test on L1 race

[AztecBot]

Summary

Fixes flake in e2e_cheat_codes > warpL2TimeAtLeastTo with target in current slot auto-adjusts to next slot observed on ARM (CI log).

The test reads currentL1Timestamp, computes target = currentL1 + 1, then calls cheatCodes.warpL2TimeAtLeastTo(target). In this composed env the production sequencer proposes blocks by warping anvil's L1 to the next slot boundary, which here is ~72 s. If that warp lands between our read and the cheat code's own internal lastBlockTimestamp() re-read, target ends up in the past and the cheat code throws is not in the future. Observed in the failure: target=1779569649, cheat-code-side currentL1=1779569720 (one full slot jump in ~4 ms wall clock).

warpL2TimeAtLeastBy(1) in the test directly above exercises the same auto-adjust path and is not affected because it computes the target inside the cheat code, so the race window between the two lastBlockTimestamp() reads is sub-ms.

Fix

Wrap the test body in a small retry loop. Each iteration re-reads lastBlockTimestamp and re-attempts the warp. Only the specific is not in the future race is retried — every other failure rethrows immediately, so we don't mask real bugs. A retry budget of 5 is plenty; the sequencer doesn't slot-jump twice in the retry window.

No production code change; the existing contract (past target throws, in-current-slot target auto-adjusts) is preserved.

Test plan

• CI green on merge-train/spartan (re-run a few times on ARM)

---

Created by claudebox · group: slackbot