Add models.dev metadata-enrichment pipeline (providers P0c-2)

Provider /models listings that return bare ids (OpenAI-style) get context
windows, thinking levels, image support, and display names from models.dev:
kigi-models owns the transform (parse_api_json — ONE field interpretation
for the bundled snapshot AND runtime refreshes), enrich_wire_model fills
gaps with wire values always winning and model availability strictly
wire-truth. Spec rows gained models_dev_id + wire_serves_metadata; all
three current platforms are wire-served, so this pipeline is provably
inert for them (byte-identical catalogs, zero egress, zero ~/.kigi writes
— adversarially verified).

Shell side: enrichment_fetch with a 24h disk cache guarded by binary
version + keep-set + future-stamp sanity (a registry change or downgrade
refetches instead of serving a catalog missing new providers), refresh of
https://models.dev/api.json filtered to registry ids, KIGI_MODELS_DEV_URL
override with case/whitespace-tolerant kill switch, fallback chain
fresh-cache > refresh > stale-cache > bundled (each step logged). The
fast path returns an empty catalog without forcing the bundled parse.

From the review: blast-radius-confined parsing (one drifted provider on
models.dev warn-skips instead of failing the whole refresh), registry-
coverage and field-coverage tests guarding script/parser drift, a path-
injectable core with 8 state-machine tests (one of which caught a guard
patch that had failed to apply), _meta provenance stamp in the snapshot,
and models.dev (MIT) attribution in NOTICE. Snapshot: 29 providers, 1124
models, 246KB, regenerated by scripts/gen_enrichment_snapshot.py (pure
filter, no transform).

crates/codegen/kigi-models/src/lib.rs

@@ -16,6 +16,8 @@
 
 use std::sync::LazyLock;
 
+pub mod enrichment;
+
 // ── Platform registry (PRD F2) ──────────────────────────────────────────────
 
 /// Env var holding the moonshot-cn API key (wins over the generic name).
@@ -82,6 +84,13 @@ struct PlatformSpec {
     console_host: Option<&'static str>,
     /// Interactive login-picker label. `None` = fall back to `display_name`.
     login_label: Option<&'static str>,
+    /// This platform's provider id on models.dev, for metadata enrichment.
+    /// `None` = not covered there (enrichment silently skips).
+    models_dev_id: Option<&'static str>,
+    /// True when the platform's `/models` listing itself serves context
+    /// window / thinking metadata — enrichment (and its network refresh) is
+    /// skipped entirely for such platforms.
+    wire_serves_metadata: bool,
 }
 
 const KIMI_CODE_SPEC: PlatformSpec = PlatformSpec {
@@ -94,6 +103,8 @@ const KIMI_CODE_SPEC: PlatformSpec = PlatformSpec {
     vendor: "Kimi",
     console_host: None,
     login_label: None,
+    models_dev_id: Some("kimi-for-coding"),
+    wire_serves_metadata: true,
 };
 
 const MOONSHOT_CN_SPEC: PlatformSpec = PlatformSpec {
@@ -109,6 +120,8 @@ const MOONSHOT_CN_SPEC: PlatformSpec = PlatformSpec {
     vendor: "Moonshot",
     console_host: Some("platform.moonshot.cn"),
     login_label: Some("Moonshot Open Platform (API key \u{b7} moonshot.cn)"),
+    models_dev_id: Some("moonshotai-cn"),
+    wire_serves_metadata: true,
 };
 
 const MOONSHOT_AI_SPEC: PlatformSpec = PlatformSpec {
@@ -124,6 +137,8 @@ const MOONSHOT_AI_SPEC: PlatformSpec = PlatformSpec {
     vendor: "Moonshot",
     console_host: Some("platform.moonshot.ai"),
     login_label: Some("Moonshot Open Platform (API key \u{b7} moonshot.ai)"),
+    models_dev_id: Some("moonshotai"),
+    wire_serves_metadata: true,
 };
 
 /// The platform registry. Platforms are compiled-in spec rows; there is no
@@ -217,6 +232,17 @@ impl PlatformId {
     pub fn login_label(self) -> &'static str {
         self.spec().login_label.unwrap_or(self.spec().display_name)
     }
+
+    /// This platform's provider id on models.dev (metadata enrichment).
+    pub fn models_dev_id(self) -> Option<&'static str> {
+        self.spec().models_dev_id
+    }
+
+    /// True when the live `/models` wire serves metadata itself — enrichment
+    /// and its network refresh are skipped for such platforms.
+    pub fn wire_serves_metadata(self) -> bool {
+        self.spec().wire_serves_metadata
+    }
 }
 
 /// Split a managed catalog key `{platform_id}/{model_id}` back into its