feat: use custom html builder for widget3

[Rathoz]

Summary

TODO: Test performance

How did you test this change?

[Copilot]

Pull request overview

This PR changes the Widget3 renderer to build HTML output via a custom string builder instead of relying on mw.html, likely aiming to reduce overhead during render.

Changes:

• Added a custom HTML string builder (buildHtmlString) including a void/self-closing tag list and attribute escaping.
• Updated Renderer.render for HTML-tag VNodes to render children and emit HTML via the new builder (with special handling for fragment).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ElectricalBoy]

how does this compare to mw.html lib performance wise?

[Rathoz]

how does this compare to mw.html lib performance wise?

It's on my TODO To test, you can test if you want. It should perform better in memory management I expect

[ElectricalBoy]

test: https://liquipedia.net/leagueoflegends/Module:Sandbox/ElectricalBoy/widget3-perf

• test1: pr version is ~0.08s slower and takes ~12KB more memory
• test2: pr version is ~0.15s faster and takes ~350KB less memory
• achievements table test (Faker): pr version takes ~16KB more memory, time difference negligible

[ElectricalBoy]

• test3: pr version is 0.3s faster but takes ~800KB more memory

[hjpalpha]

reg perf test: did you create the dev for the live comparison?
like if you use dev it likely increases memory from the imports alone if the dev modules exist compared to when they do not exist

[ElectricalBoy]

did you create the dev for the live comparison?

yes