Result session persistence: results evaporate between browser sessions

The problem

At a working meeting, a researcher described running a simulation on policyengine.org, coming back to it later, and finding it wasn't there — the result had evaporated between browser sessions. That's a reproducibility problem: the work was done, but there was no stable way to get back to it.

Why this is distinct from provenance/citation work

#940 (resolved model/data provenance) covers the long-tail answer: permalinks that resolve forever for citing published results. This issue is about the more everyday case — a researcher opens a result, navigates away, comes back, and can't recover the same numbers from that session. The fix may be simpler than full TRACE-style permalinks (e.g., a stable URL or persisted session state), and shouldn't need to wait on that larger workstream.

Likely areas to investigate

• Result/report state that doesn't survive a browser restart or session expiry.
• Whether the report URL scheme fully pins reform/baseline/dataset/period so a reload reproduces the same result.
• Client-side persistence (e.g., local storage) that isn't written or restored correctly.

Scope

Filing without a fix proposal — the first step is a concrete reproduction (the exact URL/workflow and what was expected vs. seen on return).

Migrated from policyengine-app#2833.

Start by reproducing the reported workflow on policyengine.org: record the exact result URL, then navigate away, restart the browser, and compare the returned result with what was expected. Inspect the report URL scheme and client-side persistence for whether reform, baseline, dataset, and period are preserved and restored. Done means the same result can be recovered after a browser restart or session expiry.