Retry status uploads with backoff

StatusPlugin.Status.sendToService catches and logs exceptions, then drops the report on the floor. A transient HTTP failure means the next valid status snapshot is whatever the next periodic tick produces. For ops dashboards this masks brief outages.

• Retry on failure with exponential backoff and a configurable max attempts.
• On final failure, log and skip — status reports are periodic, so unlike decision logs there's no buffer to preserve.
• Don't let retries from one tick overlap with the next scheduled tick: cap the total retry window at the upcoming interval.
• Tests: 5xx then success delivers the report; permanent failure stops retrying within the budget.

Start at StatusPlugin.Status.sendToService and trace the periodic status scheduling and existing exception logging. Add coverage for a 5xx followed by success and for permanent failure, verifying retries use exponential backoff, stop at the configured attempt and time budget, and do not overlap the next tick.