Canonical host deduplication (www.*), consolidate Nmap services and tidy reports/Graph Explorer

[Swapshadow]

Motivation

• Reduce noisy duplicate hosts such as www.X vs X when they are functionally identical, and avoid duplicated Nmap service rows in reports.
• Preserve raw data while presenting a DNSDumpster-like consolidated view for humans and downstream exports.
• Make HTML reports readable and the Graph Explorer usable by deduplicating nodes, adding display modes and filters, and making the SVG responsive.

Description

• Added a new canonicalization service backend/app/services/host_canonicalization.py that folds exact www.X aliases into X when IP/ASN/services/titles/TLS are equivalent, and produces canonical_hosts, alias_map, canonical_nmap_services and a summary object while keeping raw_hosts/raw_nmap_services.
• Integrated canonicalization into the pipeline by using it in dnsdumpster_like_inventory and the main audit flow; Nmap open ports are consolidated via consolidate_nmap_services to service_scan["canonical_open_ports"] while raw Nmap rows are preserved as raw_open_ports/raw_nmap_services.
• Updated API payloads to expose canonical_hosts, canonicalization metadata and raw fields (raw_subdomains, raw_nmap_services) and added summary counters (canonical_host_count, aliases_grouped, http_https_service_count, technology_count, etc.).
• Updated reports to use canonical data: HTML report adds an aliases column, title/TLS CN/technologies columns and a filterable A-records table; PDF and Excel sheets now show aliases and use consolidated Nmap rows while still embedding raw JSON for full detail.
• Reworked Graph Explorer backend/frontend to use canonical_hosts (store aliases in node properties), avoid separate www.* nodes when they are aliases, and added display modes (synthèse/DNS/services/technologies/risks/complète), filters, fit-to-screen/fullscreen controls and Ctrl+wheel zoom; also made the SVG responsive and reduced edge/node clutter.
• Small UX fixes: Reporting Center HTML buttons open the generated HTML in a new tab (no broken download behavior), and reports keep raw data for programmatic consumers.

Testing

• Ran Python syntax/bytecode checks with python -m compileall backend/app which succeeded (no syntax errors).
• Validated frontend JavaScript syntax with node --check backend/app/static/app.js which succeeded.
• Performed an automated canonicalization unit smoke test via PYTHONPATH=backend python - <<'PY' ... PY that exercised canonicalize_hosts on representative examples (merging www.follow... into follow..., keeping divergent www.vpn...) and asserted expected results; this test succeeded.

---

Codex Task