Account usage refresh no longer stalls on large hosted instances

Hosted Tuist’s daily account-usage refresh now completes reliably on large instances. Previously, when the backlog of stale accounts passed roughly 7,281, the daily fan-out job tried to enqueue all per-account workers in a single SQL insert that exceeded PostgreSQL’s 65,535 bind-parameter limit, causing the job to fail repeatedly. Failed inserts enqueued no work, the backlog grew every day, and the cached monthly usage rollup could stall entirely. The job now chunks inserts into batches of 1,000, so the daily refresh continues to run for accounts of any size.