feat: bundlesize dashboard

Lighthouse has already some great information on the size of all chunks into the application.

It would be nice to see those results in a dashboard where you can decide even a budget and keep track of the top 25/50/100 largest chunks, this will help detect regressions much faster.

Start by reviewing how Lighthouse currently exposes chunk-size information and how the project presents existing results. Define the dashboard, budget, and top-chunk requirements from the issue before identifying the relevant implementation and test entry points; done means regressions can be tracked against a chosen budget.

Written by the indexing model from the issue text.