Changes touching this path

• the ferry layer gets its own #288 pin, and the decision is recorded: a_keyless_reconcile_merge_does_not_resurrect_files_deleted_on_the_spine replays the deleted-on-the-spine incident in a keyless repo, so the reconcile takes the merge arm through merge_tips rather than the carry, and asserts the reconcile minted a two-parent merge so the pin cannot slide onto the carry unnoticed. restoring the pre-9763144 ancestry-union tree_at turns it red (53 passed, 1 failed in the ferry module) while the carry pin stays green and both loot-core pins go red as the positive control. the carry pin stays green there because its replay feeds the node manifest against a polluted base and carried tree, and the converge DeleteWins arm drops the stale path: with that arm also made to re-adopt theirs the carry pin goes red too (52 passed, 2 failed), and with the arm alone broken the ferry module stays green (54 passed) while four loot-core tests go red, among them delete_vs_untouched_ours_deleted_theirs_kept_deletion_wins, the carry replay input shape. carry_line doc records that reason at the seam, ADR 0028 gains a #1790 amendment, and the carry and keyless test headers drop the claims the new test made stale. no behaviour changed, only a test, comments and the ADR, and the workspace suite is green (#1790) 588d7a2b · dbf3dbe6…

Renames are not followed. loot's tree maps a path to an address, so a rename is a delete and an add. This list is the history of the name, not of the bytes.