[docs] Describe message find and tiered storage read changes in the 5.0 metrics upgrade page

[KannarFr]

Follow-up to the tiered storage read performance section: updates the 5.0 metrics upgrade page, and removes a duplicate from the OpenTelemetry metrics reference.

Metrics changes when upgrading to Pulsar 5.0.x

• The paragraph under the added OpenTelemetry instruments states that the pulsar.broker.message.find.* instruments are recorded per broker, not per topic, and what they cover.
• A new "Tiered storage reads" entry under "Existing metrics whose interpretation or exposure changes": reads that skip entries within a data block of a ledger offloaded with the jclouds offloader scan less of that block, so brk_ledgeroffloader_read_bytes_total, brk_ledgeroffloader_read_offload_rate, and the sample count of brk_ledgeroffloader_read_offload_data_latency can be lower than in 4.0.x for the same workload. These are recorded once per ranged read from the object store.

OpenTelemetry metrics reference

• The pulsar.broker.message.find.* instruments were documented twice, in "Message position search metrics" and in "Message Find Metrics". The second section is removed, and the two details only it had (when a search runs, and that pulsar-admin topics get-message-id is not included) are added to the introduction of the first.
• The link in the tiered storage overview now points to the remaining section.

The same changes are applied to docs/ and to versioned_docs/version-5.0.x/.

✅ Contribution Checklist

• I read the contribution guide
• I updated the versioned docs

🤖 Generated with Claude Code