fix: limit volume and frequency of persisted patterns

[trevorwhitney]

What this PR does / why we need it:

This PR addresses the volume of persisted patterns by delaying persistence until they are flushed from the ingesters. Previously we were persisting patterns during training. This method is ok for logs that are well structured and have few patterns, but is untenable for poorly structured logs with lots of patterns, as it produces far too much volume.

The pattern ingesters already have logic for handling high pattern churn, such as temporarily disabling detection of new patterns, and evicting infrequently used patterns. By delaying persistence until chunks are flushed, we're able to leverage all of this logic.

The downside, of course, is data durability. Patterns are stored in memory, and by default not flushed until 3hr. This PR will flush patterns on graceful shutdown, however unexpected container kills will cause data loss. Since we're just dealing with patterns that should be acceptable for now.

Checklist

• Reviewed the CONTRIBUTING.md guide (required)
• Documentation added
• Tests updated
• Title matches the required conventional commits format, see here
  • Note that Promtail is considered to be feature complete, and future development for logs collection will be in Grafana Alloy. As such, feat PRs are unlikely to be accepted unless a case can be made for the feature actually being a bug fix to existing behavior.
• Changes that require user attention or interaction to upgrade are documented in docs/sources/setup/upgrade/_index.md
• If the change is deprecating or removing a configuration option, update the deprecated-config.yaml and deleted-config.yaml files respectively in the tools/deprecated-config-checker directory. Example PR