Collected Metric was collected before with the same name and label values with MongoDB 8.3.1 #1285

Describe the bug
When using mongodb_exporter 0.51.0 against MongoDB 8.3.1 with the --collector.diagnosticdata collector enabled, the exporter logs duplicated metric registration errors related to mongodb_ss_opLatencies_transactions_histogram_micros.

The exporter appears to internally collect the same metric more than once during the gather phase, causing Prometheus registry conflicts.

The metrics endpoint itself (/metrics) does not visibly expose duplicated metrics, but the exporter continuously logs errors during collection.

To Reproduce
Steps to reproduce the behavior:

1. Run MongoDB 8.3.1
2. Start mongodb_exporter with:

mongodb_exporter \
  --web.listen-address=:9104 \
  --mongodb.uri="mongodb://127.0.0.1:27017/" \
  --collector.diagnosticdata

Observe exporter logs
The issue also happens with additional collectors enabled, but can be reproduced with ONLY:
--collector.diagnosticdata

Expected behavior
The exporter should collect metrics successfully without duplicated metric registration errors.

mongodb_ss_opLatencies_transactions_histogram_micros should either:

• be deduplicated internally before registration
• or exposed only once

Logs

time=2026-05-14T09:33:40.953Z level=ERROR source=http_error_logger.go:47 msg="error gathering metrics:" error="[from Gatherer #2] collected metric "mongodb_ss_opLatencies_transactions_histogram_micros" { label:{name:"cl_id" value:""} label:{name:"cl_role" value:""} untyped:{value:24576}} was collected before with the same name and label values" total_errors=64 error_no=58

Additional observations:

curl localhost:9104/metrics does NOT visibly show duplicated metrics
The error happens during exporter gather/registration phase
Disabling --collector.diagnosticdata removes the issue completely
--compatible-mode=false and --no-compatible-mode do not change behavior

Environment

• OS: Fedora CoreOS
• Environment: Podman Container
• MongoDB version: 8.3.1
• mongodb_exporter version: 0.51.0

Additional context
The issue appears specific to MongoDB 8.3.x telemetry changes and the diagnosticdata collector.

The problem can be reproduced with only:

--collector.diagnosticdata

No other collectors are required to trigger the issue.

The metrics endpoint still responds, but the exporter continuously logs gather errors.