docs: add single-host monitoring guide for Arc EL + CL

[Bela1401]

Summary

This PR adds a practical monitoring guide for operators running a single Arc
execution layer + consensus layer on one Ubuntu host.

It documents:

• Prometheus scrape targets for EL and CL metrics
• a minimal Grafana + Prometheus setup
• dashboard provisioning
• local access via SSH tunneling
• common troubleshooting notes observed during real-world setup

Scope

This PR intentionally stays focused on a single-host operator setup and does not
attempt to cover multi-node or production observability architectures.

Related docs change

This PR also adds a reference from docs/running-an-arc-node.md to the new
docs/monitoring.md guide.

Notes

This PR supersedes the earlier closed PR #28, which was auto-closed after the
target branch changed.

[Bela1401]

Reopened this as a fresh PR on top of the updated main, as suggested in the previous discussion.

This supersedes #28.

[Bela1401]

Validation screenshots from the earlier single-host setup are attached below for context.

[sdaveas-circle]

I tested the guide's configs against a running EL+CL network (Quake testnet with 5 validators). The technical content is solid:

• Prometheus scrapes both EL (/) and CL (/metrics) successfully
• Grafana auto-provisions 8 dashboards from the repo's existing JSONs
• Datasource UID prometheus resolves correctly
• Real metrics data flows end-to-end (verified reth_info, arc_malachite_app_block_time_count, both targets up)

The configs, ports, paths, and network_mode: host approach all work correctly. One critical formatting issue blocks merging — see inline comment.