perf(interrupt-rotation): parallelize slow sections with fragments

Previously the page fetched the full open issue/PR snapshot synchronously at
the top of the script and fetched the monitored-repo PRs inline mid-page, so
both blocked the main thread in series before the existing parallel fragments
were even dispatched. The heavy CI-artifact downloads never overlapped with
the issue/PR fetch.

Move every slow data source into a parallel=True fragment and reduce the main
body to fast chrome (title, sidebar, header, dividers). The scattered action
items are grouped into two fragments that share the cached issue/PR builder,
so the snapshot is still fetched once but now overlaps with the CI metrics,
flaky tests, monitored-repo PRs, reported bugs, and repro checks. Wall-clock
drops from the sum of the blocking fetches to roughly the slowest single one.

Visual order, help text, and column configs are unchanged.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: sfc-gh-bnisco