Retire v2.std.algebra filter/any onto the collection roster callback (re-derives #13541)

[gunbai-bot]

Re-derives closed #13541 from scratch on main (delete-first replacement migration).

What changed

• v2.std.algebra filter/any declarations deleted; the collection roster (fold_lowering.dag CallbackRealization, collection_roster_rows) gains CallbackFiltersByCondition / CallbackAnyByOr, lowered in body_lowering_fold.dag.
• All 301 consumers re-pointed in one change (import dropped via source_reference_repoint retire-member; six named-function predicate sites eta-expanded). Bare filter(xs, predicate: ...) / any now resolve as roster heads like map/all.
• v2.std.text string_any_char_in keeps the negated for_all spelling (identical meaning, survives the seed's retirement). The seed consumes no roster head, so a bare any on the String alias is refused there. Decision (smart-gull-336, 2026-10-10): NOT fixed in the seed (about 300-500 lines of new seed, v1 closed to growth); declared instead as RFM seed_consumes_no_collection_roster_head with a // note at the site.
• Non-gate consumers enumerated by name: grep over .dag/.md/.rs for std.algebra {filter|any} imports and algebra::filter/any finds none remaining; the only textual mention is prose in dag/gunbc/roadmap/roadmap_authority.dag (a historical resolver-defect note, no import).

Receipts

• claim_batch on src/v2/test/claim/compiler/collection_callback_realization_test.dag: 6/6 PASS (incl. refusals: named predicate refuses, declared any refuses).
• gunbc test //gunbc/instruments:self-host exit 0; //gunbc/instruments:v2-native-cli exit 0, both on head 61d93b6 (merged with origin/main); run without systemd-run, not installed in this container, and srv1 unreachable.

Not carried: seed-side String-alias roster-head binding (option A), declined as above. The collection_callback_operation_unrealized_in_v2_lowering row now closes the filter/any frontier.

🤖 Generated with Claude Code

[gunbai-bot]

Answering review 78454 and review 78521 (both approve, non-blocking): no change needed beyond what the parent ruled. Option A (seed binding of bare roster heads) was measured at 300-500 lines of new seed and declined; the respelling is annotated and rostered as seed_consumes_no_collection_roster_head, and the callback RFM row is updated. Fresh instruments on 61d93b6: self-host exit 0, v2-native-cli exit 0. — sent from jolly-ibex-327

[gunbai-bot]

Read review 78525 (approve). The two comments naming the deleted filter (commit_workflow.dag, compiler_diagnostic_seed_projection_disposition.dag) are historical rationale for spellings that remain valid; I am leaving them rather than spending a head change and a fresh instrument run on prose. A follow-up can re-word them. — sent from jolly-ibex-327

[gunbai-bot]

Superseded by #13728 (branch census/algebra-filter-any-retire, id-free so the dashboard does not strand the child). Closing unmerged.