scan: introduce group pages

[bagder]

For different views/collection of data points next to each other

[Copilot]

🟡 Changes recommended

A typo in the %groups keys prevents CPU request graphs from ever being grouped, and the new group file writes should check for open() failures to avoid silently incomplete output.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR introduces “group pages” that aggregate selected performance graphs into shared HTML pages, enabling alternative views/collections of related data points alongside the existing per-test rendering.

Changes:

• Add a %groups mapping to associate specific test graphs with one or more group pages.
• Append each rendered test’s “most recent rounds” graph image into the relevant group HTML files during show().
• Generate initial group page files and add links to them from the main index output.

File summaries

File	Description
scan.pl	Adds group definitions, writes/updates per-group HTML pages during rendering, and links group pages from the index output.

Review details

Suppressed comments (3)

scan.pl:44

• The group mapping key has an extra '"' character, so lookups for the "h2-req-cpu" test will never match and the CPU graph won't be added to the group pages.

    'h2-req-cpu"' => 'http2,cpu',

scan.pl:54

• The group mapping key has an extra '"' character, so lookups for the "h3-req-cpu" test will never match and the CPU graph won't be added to the group pages.

    'h3-req-cpu"' => 'http3,cpu',

scan.pl:1158

• Creating the initial group files doesn't check open() for failure; if this fails, later appends in show() will either fail or append to stale content from a previous run.

for my $g (keys %groupnames) {
    open(G, ">$outdir/$g.html");
    print G "<a href=\"index.html\">Back to perf index</a>\n";
    close(G);

• Files reviewed: 1/1 changed files
• Comments generated: 2
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.