Reduce ENS data cache TTL from 1 week to 1 day

Nobody has claimed this yet.

## Problem

All three ENS API routes use `getCacheControlHeader("week")` which sets `s-maxage=604800` (7 days) + 14 days `stale-while-revalidate`. When an orchestrator or gateway updates their ENS name, description, or avatar, changes can take up to 7–21 days to appear on the Explorer.

## Proposed Change

Change `getCacheControlHeader("week")` to `getCacheControlHeader("day")` in these three files:

- `pages/api/ens-data/[address].tsx` (line 22)
- `pages/api/ens-data/index.tsx` (line 18)
- `pages/api/ens-data/image/[name].tsx` (line 51)

This gives `s-maxage=86400` (1 day + jitter) with 7-day `stale-while-revalidate`, matching the original setting before it was bumped to `"week"` in commit 40e0fe2.

## Context

The week-long cache was introduced in #216 as a cost/performance optimization to reduce L1 ENS RPC calls. A 1-day cache still provides significant caching while reducing the staleness window for orchestrator/gateway identity updates.

Start with the three ENS API route files named in the issue: pages/api/ens-data/[address].tsx, pages/api/ens-data/index.tsx, and pages/api/ens-data/image/[name].tsx. Check each cache-control call and verify that all three use the day setting, producing a one-day s-maxage and the described stale-while-revalidate window.

Written by the indexing model from the issue text.