fix: guard against failed ingester requests for detected fields

[trevorwhitney]

What this PR does / why we need it:

We have seen an increase in 500s from ingester for various queries. These 500s are causing entire detected field requests to fail. Since detected fields is just an approximation/sample anyway, this PR changes that behavior to guard against that in a few ways

1. split requests in the querier into separate ingester and store requests, and process iterators from either if any are returned
2. return an empty response rather than an error when no iterators are returned (we still log the error as a warn). this will give other splits in the QF a chance of returning data even if not all subqueries are successful.
3. Remove the retry middleware for /detected_fields as it's just an approximation anyway, it doesn't need every subquery to return
4. Add a fall through middleware so that failed request return an empty result rather than an error.

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
• For Helm chart changes bump the Helm chart version in production/helm/loki/Chart.yaml and update production/helm/loki/CHANGELOG.md and production/helm/loki/README.md. Example PR
• 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

[trevorwhitney]

closing in favor of #14212