Fix duplicate index throttling metric registration

[wdtte]

Summary

• rename the index-level throttling metric to elasticsearch_index_stats_indexing_is_throttled
• avoid the registry collision with the node-level elasticsearch_indices_indexing_is_throttled metric
• add a regression test to ensure NewNodes and NewIndices can be registered together

Reproduction

The issue was reproduced in multi-target /probe mode with these flags:

• --es.uri=
• --config.file=/etc/elasticsearch-exporter/exporter-config.yml
• --web.listen-address=:9114
• --es.all
• --es.shards
• --es.ssl-skip-verify

Representative runtime logs:

time=2026-05-26T12:48:36.027Z level=INFO source=tls_config.go:354 msg="Listening on" address=[::]:9114
time=2026-05-26T12:48:36.027Z level=INFO source=tls_config.go:357 msg="TLS is disabled." http2=false address=[::]:9114
2026/05/26 12:49:02 http: panic serving 10.34.29.6:36890: a previously registered descriptor with the same fully-qualified name as Desc{fqName: "elasticsearch_indices_indexing_is_throttled", help: "Whether indexing is currently throttled for an index (1=throttled, 0=not throttled)", constLabels: {}, variableLabels: {index,cluster}} has different label names or a different help string
github.com/prometheus/client_golang/prometheus.(*Registry).MustRegister(...)
	/go/pkg/mod/github.com/prometheus/client_golang@v1.23.2/prometheus/registry.go:406
main.main.func3({0xdc6940, 0x298a9f4b0690}, 0x298a9f3fb900)
	/app/main.go:403 +0xe3c

The same panic repeats on every scrape while NewNodes() and NewIndices() are both registered.

Testing

• env GOCACHE=/Users/wdte/GolandProjects/elasticsearch_exporter/.gocache go test ./collector -run TestNodesAndIndicesRegisterTogether -count=1

Notes

• full go test ./collector/... is blocked in the current sandbox because existing tests use httptest.NewServer and local port binding is not permitted in this environment

[wdtte]

This PR is ready for review.

It fixes a reproducible /probe panic caused by duplicate registration of elasticsearch_indices_indexing_is_throttled when NewNodes() and NewIndices() are both registered.

The change is intentionally small:

• rename the index-level metric to elasticsearch_index_stats_indexing_is_throttled
• keep the node-level metric unchanged
• add a regression test for co-registration

DCO is green now. A maintainer review would be appreciated.

@sysadmind could you please take a look when you have time?

This seems to be a regression caused by the interaction between multi-target /probe registration and the index-level indexing metrics added in #1125. I added a minimal fix plus a regression test.

If this is merged, an updated official image would be very helpful, since I am currently hitting this issue in production.