fix(e2e): pause L1 mining during slot query + warp in epochs_ha_sync

[spypsy]

Fixes a ~1.7% flake in epochs_ha_sync caused by a race between Anvil's interval mining and the test's slot computation + warp sequence. During the ~8s spent proving/sending transactions, L1 blocks kept being mined, and by the time warp() was called, the chain had already advanced past the target timestamp. Wraps the slot query + warp in execWithPausedAnvil so L1 can't advance between the two calls.

Fixes A-892

[spalladino]

I'm not sure this makes a difference. The slot we're jumping ahead to is 2 slots in the future, computed from the slot number we've just read. There's no proving/sending txs in between, so there's no way that 8s pass between the read and the warp.

By the way, does this flake still happen? I had tried fixing it last week in this PR, by jumping 2 slots instead of 1, to prevent the race condition of 1) reading the current slot, 2) time passes and we move into it, 3) we try warping into it.