Speed up first sync of the EVM-reader for new applications

[vfusco]

Context / Problem

At first startup the EVM-reader issues an eth_getLogs filter-log scan from the block where the InputBox contract was deployed all the way to the latest L1 block.
For a brand-new dApp this huge fromBlock … toBlock range contains hundreds of thousands of empty blocks, so the node burns JSON-RPC quota and spends minutes waiting for results even though no relevant logs exist.

---

Suggested Solution

1. Fast “probe-and-decide” flow
   Let
   current = latest L1 block,
   appDeploy = dApp contract deployment block,
   inputDeploy = InputBox deployment block.

   Step A – probe current

   • Call eth_getLogs for this dApp only at current.
     • No inputs: set last_checked_block = current and finish (zero-cost start).
     • Some inputs: continue to Step B.

   Step B – probe appDeploy − 1

   • Call eth_getLogs at block appDeploy − 1.
     • No inputs: sync the shorter range [appDeploy … current].
     • Some inputs: fall back to the full scan [inputDeploy … current].

   The optimisation costs at most two extra RPC calls.

2. State & restart safety
   Persist last_checked_block in the DB so restarts reuse the determined range.

---

Acceptance Criteria

#	Scenario	Expected result
1	Brand-new dApp, no inputs	Sync finishes after ≤ 2 RPC calls; node ready in < 2 s.
2	Inputs only after deploy	Scans [appDeploy … latest]; RPC/time drop ≥ 80 %.
3	Inputs before deploy	Falls back to full scan; behaviour identical to current.

This optimisation slashes first-run synchronisation time and RPC usage for new applications while preserving correctness for existing ones.