Add scheduled metrics cleanup job

[neddp]

What is this change about?

This change adds automatic cleanup of stale metrics files in /var/vcap/store/director/metrics to prevent disk space issues caused by accumulating Prometheus DirectFileStore binary files.

Fixes #2623

Please provide contextual information.

Background:

• Issue Stale metrics-server files in /var/vcap/store/director/metrics #2623: Stale metrics-server files accumulating in /var/vcap/store/director/metrics
• Issue Metrics server /api-metrics endpoint fails with 500 Internal Server Error without helpfull diagnostic #2332: Years-old binary files (metric_*.bin) were causing Prometheus scraping to fail
• Directors that run for years can accumulate hundreds/thousands of these files

Implementation:

• Added new ScheduledMetricsCleanup job that runs daily at midnight UTC
• Configurable retention period (default: 7 days)
• Only runs when director.metrics_server.enabled is true
• Follows existing BOSH patterns (similar to orphaned disk/VM cleanup jobs)

What tests have you run against this PR?

• RuboCop linting
• Unit tests

How should this change be described in bosh release notes?

New Feature: The metrics server now automatically cleans up stale metrics files older than the configured retention period (default: 7 days). This prevents disk space issues from accumulating metric binary files. Cleanup runs daily at midnight UTC and can be configured via director.metrics_server.file_retention_days and director.metrics_server.cleanup_schedule properties.

Does this PR introduce a breaking change?

No. This is a new feature with safe defaults:

• Only runs when metrics_server is enabled
• Default retention is 7 days
• Cleanup can be disabled by setting file_retention_days to 0
• No changes to existing behavior or APIs

Tag your pair, your PM, and/or team!