Update RFC 086 to reflect what was actually built

[rcantin-w]

• For IVR: Post release, remove toggle and cleanup wellcomecollection.org#12990

What does this change?

RFC 086 still said "Status: Not started" with every phase unticked, despite phases 0-6 having shipped. More to the point, the phase documents were written before implementation and use names the code never adopted - ItemViewerContextV2, the .legacy.tsx/.refactored.tsx filename suffixes, and an iiifViewerRefactored toggle. None of those appear anywhere in wellcomecollection.org. Phase 7 is the only phase still outstanding, and as written every path in its mv/rm commands is fictional.

• README.md - status and last-modified updated, phases 0-6 ticked with links to their tickets, plus a table translating the provisional names to the real ones (contexts/ItemViewerContext/refactored.tsx, legacy//refactored/ directories, itemViewerRefactor). Also fixes a mangled principle 8/9 and a missing blank line.
• 13-phase-7-cleanup.md - steps rewritten against the actual layout so they can be followed. Adds three the original missed: collapsing the context barrel, getting every consumer onto one import path, and unpicking the dual-context test harness in render.tsx. Fixes a broken link to 15-risks-and-success.md.
• 12-phase-6-duplicate-calls.md - records that this phase became an audit rather than the planned find-and-replace, what it turned up (one real duplicate in GridViewer), and the four lookalikes deliberately left alone.
• 06 through 11 - each phase's success criteria checked off against the code.
• 13-migration-checklist.md - phase 6 and 7 rows brought in line.
• rfcs/README.md - regenerated by the pre-commit hook, since 086 is now the most recently modified RFC.

Three criteria are left unticked because they genuinely weren't met, each annotated with what happened instead:

• Phase 1's "no console warnings" - a temporary log reporting which context is in use was added deliberately. Removing it is item 1 of IVR: Topics to review wellcomecollection.org#13273.
• Phase 5's "restricted badge appears correctly" - there is no restricted badge in either tree. What isCurrentCanvasRestricted actually gates is hiding the download button on a restricted canvas.
• Phase 6's "all components use currentCanvasIndex from context" - that value was deliberately never added to the context, since it would duplicate query.canvas.

Two more are left unticked because the code can't evidence them: phase 0's "all team members understand what types exist", and phase 3's optional manual testing checklist. Each phase's separate manual-testing checklist is also still unticked, for the same reason.

The eight documents that mention the old provisional names are otherwise left alone, as records of phases already delivered; the README table covers reading them.

How to test

Read it. All relative links resolve, and every ticket number was checked against the GitHub API.

Have we considered potential risks?

Docs only, nothing deployed. The risk runs the other way: phase 7 is the one part anyone still has to follow, so if its rewritten steps are wrong, someone hits that at cleanup time - worth a careful read of that file specifically. The counts in step 7.5 (32 files importing the context barrel, 16 importing refactored directly) are from today's main and will drift, so they're written as a grep to re-run rather than a number to trust.

Written by Claude Code.

[davidpmccormick]

Neat-oh!

[davidpmccormick]

Suggest linking to the issue (#12988) instead of the PR (#13412) for consistency?

[rcantin-w]

@copilot can you do that? (I've no idea I'm trying it)

[Copilot]

Done in b6f0fc0 — phase 5 now links to issue #12988 instead of PR #13412.

[rcantin-w]

It did it.