variant_field_lowering: five claims read what normalize emits now (source-spelled Int, the wildcard form)

[gunbai-bot]

Red: this was reported by wise-bat-862 as 4 failing claims in v2.test.parse.variant_field_lowering, predating #12629. Reproduced with claim_batch on main.

• On main before XL-2: a wildcard pattern lowers to its own form, not an atom spelled _ #12597 landed: 4 FAIL.
  • vf_positional_variant_declares_one_required_field_zero
  • vf_braced_variant_declares_named_field
  • vf_record_declaration_is_named_with_declared_fields
  • vf_shadowing_binder_is_carried_on_the_arm
• On main after XL-2: a wildcard pattern lowers to its own form, not an atom spelled _ #12597 landed today: vf_wildcard_pattern_keeps_its_field also fails.

Why nobody saw it: the module v2.test.parse.variant_field_lowering matches no selector in v2.workflow.required_floor required_gate_admits, so the required floor declines it before execution. Both changes landed green while reddening it. This is the population of the required_gate_bankruptcy drop ("every witness outside the roster"), and its restoration trigger already covers it. I'm not adding a new row.

Chain (DESIGN §6b). The subject is normalize: source in, normalized tree out. The earliest unjustified boundary is each claim's expectation, not the lowering.

• v2 resolve: a kernel type spelling is read by its declaration (own Int shadows, imported user Int refuses) #12540 moved the kernel binding of a type spelling from body lowering to resolve, after the scope walk, so a module's own type Int can shadow it (the silent-wrongness fix recorded as kernel_type_spelling_captured_a_module_declaration). Normalize therefore now emits Int as written, and the 4 claims were asserting a normalize output that is now resolve's job. The kernel binding itself stays asserted at resolve by v2.test.claim.body_let_annotation.
• XL-2: a wildcard pattern lowers to its own form, not an atom spelled _ #12597 gave an authored _ its own unauthorable form, read by the one recogniser v2.std.node_query is_wildcard_pattern. The wildcard claim now asks that recogniser instead of matching an atom spelled _.

Evidence (local claim_batch):

• With this change: all 34 claims in the module PASS, including the 17 vfs_fidelity_* controls.
• Controls: the old expectations are the measured reds above (4 on pre-XL-2: a wildcard pattern lowers to its own form, not an atom spelled _ #12597 main, 5 on current main).

🤖 Generated with Claude Code