Track fix: material dataset hourly refresh cron not running on hosts

Nobody has claimed this yet.

## Problem
The hourly material dataset refresh is not running because the cron job is not executing on hosts.

## Why this matters
If refreshes are missed, dataset freshness degrades and downstream discovery behavior may become stale.

## Scope
Restore reliable hourly refresh execution across hosts and add detection for missed runs.

## Tasks
- Audit host cron state (service enabled/running, user crontab, deployed entry)
- Validate job command, environment, and permissions in cron context
- Inspect cron/job logs to identify failure mode
- Fix deployment/runtime configuration so job executes hourly on all hosts
- Run one manual/backfill refresh and verify updated freshness markers
- Add alerting when refresh has not succeeded in >1 hour

## Acceptance criteria
- Job runs successfully every hour on all hosts
- Success is visible in logs/metrics
- Missed-run alerting is active

Start by auditing host cron service state, user crontabs, deployed entries, and the job's cron-context command, environment, and permissions. Inspect cron and job logs to identify the failure mode, then run a manual or backfill refresh and verify freshness markers. Done means hourly execution works on all hosts, success appears in logs or metrics, and missed runs trigger an alert after more than one hour.

Written by the indexing model from the issue text.