Workspace INDEX.md + flag §07 DI non-conformance

Two changes:

1. INDEX.md (new, generated). Workspace-level index that lists every
   package grouped by UAPF level (L0/L1/L2/L3/L4) for navigation. Per
   UAPF specification §01-concepts.md, levels are an aggregation and
   governance scope only and MUST NOT be used to imply modeling
   semantics; per §04-folder-structure.md the SHOULD-recommended
   on-disk layout is enterprise/ + domains/ + processes/ regardless of
   level. INDEX.md is therefore a level-grouped *view* over the
   spec-conformant on-disk layout, not an alternative layout. Generated
   by tools/build-index/build_index.py (new), which walks the workspace
   manifests and emits INDEX.md from current state.

2. docs/methodology.md updated to document the §07 auto-layout DI as a
   known deliberate non-conformance. §07-package-format.md requires
   authored DI (from a conforming OMG modeler) and explicitly forbids
   automatic layout generation as a substitute. The DI in the five
   .bpmn files in this workspace is auto-generated by
   tools/register-transcoder/bpmn_di.py and is therefore non-conformant
   under §07. It is kept for the POC so artefacts preview visually;
   the conformant path is to re-author each .bpmn in Camunda Modeler
   or bpmn-js Studio and recommit, which emits authored DI. New
   *Known non-conformances* section in methodology.md cites §07
   verbatim and explains this rationale. The Pass-1 transcoder section
   and Final-validation-pass section are softened to call out that
   the DI is auto-generated.

uapf-cli validate green on all 13 packages (L1 domains/gramatvediba,
L2 fg1-fg6, L3 fg3-2/3/6, L4 fg3-1/4/5).

docs/methodology.md

@@ -114,10 +114,14 @@ one `bpmn:startEvent` per *entry step* (no in-group predecessor) and one
 `bpmn:endEvent` per *exit step* (no in-group successor), so the fragment's
 real boundary is visible rather than hidden behind synthesised gateways.
 The output then has BPMN Diagram Interchange (`bpmndi:BPMNDiagram` with
-`BPMNShape` and `BPMNEdge` elements) appended by `bpmn_di.py` using a
+`BPMNShape` and `BPMNEdge` elements) appended by `bpmn_di.py` via a
 swim-lane left-to-right auto-layout, so the resulting file previews in
-bpmn.io, Camunda Modeler and the ProcessGit web view without manual
-positioning.
+bpmn.io, Camunda Modeler and the ProcessGit web view. **This DI is
+auto-generated, not authored** — §07 of the UAPF specification requires
+authored DI produced by a conforming OMG modeler and explicitly forbids
+automatic layout generation as a substitute. The auto-layout output is
+kept as a known deliberate non-conformance pending re-authoring of each
+`.bpmn` in a modeler. See *Known non-conformances* below.
 
 The output is `isExecutable="false"` and deliberately unembellished: no
 inferred gateways, no synthesised decision logic, no compensation for
@@ -218,8 +222,9 @@ correction (next section):
   `sourceRef`/`targetRef` nodes; every `flowNodeRef` resolves to a defined
   node; every `incoming`/`outgoing` reference is consistent with the
   corresponding flow's source/target.
-- All `.bpmn` files now carry BPMN Diagram Interchange — they preview
-  cleanly in bpmn.io, Camunda Modeler and ProcessGit's web view.
+- All `.bpmn` files carry BPMN Diagram Interchange (auto-generated; see
+  *Known non-conformances* below) — sufficient for preview in bpmn.io,
+  Camunda Modeler and ProcessGit's web view.
 - The transcoder is byte-deterministic: re-running it on the FG3 register
   for 3.5.2 and 3.5.3 reproduces the committed `sample-output/` files
   exactly.
@@ -249,6 +254,31 @@ packages may reference L4 directly when no intermediate composition is
 needed (`fg3` `includes` references the three L4s and the three L3 stubs
 in parallel, which is conformant).
 
+## Known non-conformances
+
+This POC ships with one documented non-conformance against the UAPF
+specification, retained deliberately for the POC and tracked here.
+
+**Authored Diagram Interchange (§07-package-format.md).** §07 makes DI a
+MUST for cornerstone BPMN/DMN/CMMN files and explicitly forbids automatic
+layout generation as a substitute:
+
+> "An implementation MUST NOT rely on automatic layout generation as a
+> substitute for authored DI: a generated layout is not the authored
+> layout and is not deterministic across tools."
+
+The DI in the five `.bpmn` files (`fg3-1`, `fg3-4`, `fg3-5`, and the two
+transcoder samples under `tools/register-transcoder/sample-output/`) is
+produced by `tools/register-transcoder/bpmn_di.py` using a swim-lane
+left-to-right auto-layout. It renders the diagrams in bpmn.io, Camunda
+Modeler and ProcessGit's web view, but it is not authored DI and
+therefore non-conformant under §07. The auto-layout is kept so the POC
+artefacts are reviewable visually; the conformant path is to open each
+`.bpmn` in a conforming OMG modeler (Camunda Modeler, bpmn-js Studio) and
+re-save, which emits authored DI. The DMN files are unaffected — §07
+exempts DMNs whose only logic is decision tables, and all three workspace
+DMNs are exactly that.
+
 ## Implications for the AI regulatory sandbox
 
 The pipeline has four properties that bear on the sandbox's evaluation.