Slate v2 issue claims need exact browser proof and honest input contracts

Problem

Slate v2 issue ledgers can drift if a row is promoted from source ownership alone. Browser-facing issues need a browser row that follows the original user flow, and input-event issues need an honest statement of whether Slate promises native event parity or a model-owned editing callback.

Symptoms

• #5826 had selection repair and scroll forwarding code, but no exact long-editor blur, scroll, and refocus proof.
• #5603 and #5669 asked for native input event parity, while the runtime intentionally routes text through beforeinput and Backspace through a model-owned keydown command.
• bun test packages/slate-history/test/history-contract.ts failed with "filters did not match any test files"; Bun needed ./packages/... to treat it as a path.

What Didn't Work

• Treating scroll-selection source tests as enough for #5826. The issue was about viewport movement after a concrete long-editor refocus flow.
• Treating model-owned input tests as native input parity. That would overstate the contract and make the issue ledger less trustworthy.
• Running Bun's test path without ./, which triggered filter semantics instead of path semantics.

Solution

Write the smallest browser rows that match the issue claims:

• For #5826, add a huge-document row that clicks the top block, blurs the editor, scrolls to the final block, clicks back into the editor, and asserts the clicked final-block selection stays visible.
• For #5603 and #5669, add a richtext row that proves the actual contract: start and number insertions are observable through beforeinput, while Backspace is a model-owned keydown delete command. Keep native input parity unclaimed.
• For package-path Bun tests, use bun test ./packages/....

Then sync all claim surfaces in the same pass:

• docs/slate-v2/ledgers/issue-coverage-matrix.md
• docs/slate-v2/ledgers/fork-issue-dossier.md
• docs/slate-issues/gitcrawl-v2-sync-ledger.md
• docs/slate-v2/references/pr-description.md
• the active docs/plans/** execution plan
• the session active goal state

Why This Works

Issue claims stay credible because every promotion has the same shape as the reported user path. #5826 can become Fixes because the browser proof matches the long-editor refocus report. #5603 and #5669 stay Related because the proof clarifies Slate's current input contract without pretending native input events are guaranteed for every operation.

Prevention

• Promote a Related issue only when the browser row follows the issue's original steps closely enough that a maintainer would recognize the repro.
• Name the contract honestly. If Slate owns the operation through beforeinput or keydown, do not call that native input parity.
• After any issue classification change, update the coverage matrix, fork dossier, manual sync ledger, PR reference, active plan, and completion state before marking the lane done.
• Use bun test ./path/to/file.ts for direct Bun test-file execution.

Related Issues

• #5826
• #5603
• #5669
• #3871
• #5847
• #3534
• #3551