5-test-migration (3): delete v1 node_minimal_representation_test — floor witnesses already cover it

[briansrls]

Third PR of the 5-test-migration lane (v1 burn-down). Net diff vs main is pure deletion — git diff origin/main -- src/v2 dsl is empty (the in-branch perturb commit is reverted; squash-merge flattens it).

What

Delete src/v1/tests/src/node_minimal_representation_test.rs (149 LOC, 1 #[test]) and its mod line in lib.rs.

Why the floor fully covers it

The v1 module was a duplicate harness: it resolved src/v2/test/claim/manual/node_minimal_representation_test.dag and invoked 9 of its 10 test fn witnesses through the v1 interpreter, asserting Bool(true) for each. All 10 test fns are floor-enrolled: the file is under the src/v2 witness layer root (*_test.dag marker discovery), and src/v2/test/claim/manual is additionally a live witness_discovery_scan_dirs entry (cli_run.rs authority projection). Floor coverage is a strict superset of the v1 module's 9 invocations. No hand-enrollment added.

Receipts (green-by-execution + perturb→red)

1. Green: claim_batch --source-root src/v2 --source-root dsl --entry src/v2/test/claim/manual/node_minimal_representation_test.dag --functions <all 10> → 10/10 PASS (1 resolve 1546ms, exit 0).
2. Discriminating (perturb→red): temporarily committed a subject perturbation (drop NsfTypeAnnotation from minimal_kind_owed_placement in src/v2/std/node_minimal.dag, commit 1cda9ceb9, reverted in-branch):
   • witness_partition_covers_superset_holds → FAIL
   • node_minimal_owed_bucket_count_holds → FAIL
   • node_minimal_kind_of_expr_leaf_holds → PASS (unrelated control), exit 1.
3. cargo test -p v1-compiler-tests --no-run compiles clean after the delete.

Mapping note

Everything the v1 module asserted (clean resolve of the entry; 9 witnesses true) is subsumed by floor execution of the same witnesses; the 10th test fn (node_minimal_roster_length_matches_superset_count_holds) was floor-only already. Nothing remains un-migrated for this module.

🤖 Generated with Claude Code