feat: media expansion surfaces — edit mask, upscale, background removal, interpolation, diarization, meshgen (ADR-0020)

- imagegen.EditRequest.Mask -> sd-server img2img inpainting (white=repaint)
- imagegen.Upscaler + BackgroundRemover, videogen.Interpolator,
  audio.DiarizationModel: new optional provider-minted surfaces
- NEW meshgen leaf package (image->3D, glb/stl/obj)
- provider/llamaswap: all five via the /upstream/<model>/<path> passthrough
  (upstreamPath helper, shared one-file multipart builder); binary success
  bodies validated (non-image, non-video, JSON-mesh rejection); diarization
  pins output=json (vtt/srt drop speaker labels)

Co-Authored-By: Claude Fable 5 <[email protected]>

provider/llamaswap/mesh_diarize_test.go

@@ -0,0 +1,170 @@
+package llamaswap
+
+import (
+	"context"
+	"encoding/base64"
+	"encoding/json"
+	"errors"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"gitea.stevedudenhoeffer.com/steve/majordomo/audio"
+	"gitea.stevedudenhoeffer.com/steve/majordomo/llm"
+	"gitea.stevedudenhoeffer.com/steve/majordomo/meshgen"
+)
+
+func TestMeshGenerate(t *testing.T) {
+	png := pngFixture(t)
+	stl := []byte("solid mort\nendsolid mort\n")
+	var gotPath string
+	var gotBody map[string]any
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		gotPath = r.URL.Path
+		_ = json.NewDecoder(r.Body).Decode(&gotBody)
+		w.Header().Set("Content-Type", "application/octet-stream")
+		_, _ = w.Write(stl)
+	}))
+	defer srv.Close()
+
+	p := New(WithBaseURL(srv.URL), WithHTTPClient(srv.Client()))
+	mm, err := p.MeshModel("image3d-hunyuan21")
+	if err != nil {
+		t.Fatalf("MeshModel: %v", err)
+	}
+	res, err := mm.Generate(context.Background(),
+		meshgen.Request{Image: meshgen.Image{MIME: "image/png", Data: png}},
+		meshgen.WithFormat("stl"), meshgen.WithSeed(7))
+	if err != nil {
+		t.Fatalf("Generate: %v", err)
+	}
+	if gotPath != "/upstream/image3d-hunyuan21/generate" {
+		t.Errorf("path = %q", gotPath)
+	}
+	if gotBody["type"] != "stl" || gotBody["image"] != base64.StdEncoding.EncodeToString(png) {
+		t.Errorf("type/image = %v/(b64 mismatch)", gotBody["type"])
+	}
+	if gotBody["seed"] != float64(7) {
+		t.Errorf("seed = %v", gotBody["seed"])
+	}
+	if _, ok := gotBody["octree_resolution"]; ok {
+		t.Error("octree_resolution sent when unset; want omitted")
+	}
+	if res.Mesh.Format != "stl" || res.Mesh.MIME != "model/stl" || string(res.Mesh.Data) != string(stl) {
+		t.Fatalf("mesh = %+v", res.Mesh)
+	}
+}
+
+func TestMeshGenerateRejectsJSONBody(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "application/octet-stream")
+		_, _ = w.Write([]byte(`  {"detail": "queue full"}`))
+	}))
+	defer srv.Close()
+
+	p := New(WithBaseURL(srv.URL), WithHTTPClient(srv.Client()))
+	mm, _ := p.MeshModel("image3d-hunyuan21")
+	_, err := mm.Generate(context.Background(),
+		meshgen.Request{Image: meshgen.Image{Data: pngFixture(t)}})
+	var apiErr *llm.APIError
+	if !errors.As(err, &apiErr) {
+		t.Fatalf("err = %v, want APIError for JSON success body", err)
+	}
+}
+
+func TestMeshGenerateRejectsBadArgs(t *testing.T) {
+	p := New(WithBaseURL("http://unused"))
+	mm, _ := p.MeshModel("image3d-hunyuan21")
+	if _, err := mm.Generate(context.Background(), meshgen.Request{}); !errors.Is(err, llm.ErrUnsupported) {
+		t.Errorf("no image: err = %v, want ErrUnsupported", err)
+	}
+	if _, err := mm.Generate(context.Background(),
+		meshgen.Request{Image: meshgen.Image{Data: []byte{1}}, Format: "fbx"}); !errors.Is(err, llm.ErrUnsupported) {
+		t.Errorf("format fbx: err = %v, want ErrUnsupported", err)
+	}
+}
+
+func TestDiarize(t *testing.T) {
+	var gotPath string
+	var gotQuery map[string][]string
+	var gotField string
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		gotPath = r.URL.Path
+		gotQuery = r.URL.Query()
+		if err := r.ParseMultipartForm(1 << 20); err != nil {
+			t.Errorf("parse multipart: %v", err)
+		}
+		if f, _, err := r.FormFile("audio_file"); err == nil {
+			gotField = "audio_file"
+			f.Close()
+		}
+		_, _ = w.Write([]byte(`{
+			"text": "hello there. general kenobi.",
+			"language": "en",
+			"segments": [
+				{"start": 0.0, "end": 1.2, "text": "hello there.", "speaker": "SPEAKER_00"},
+				{"start": 1.4, "end": 3.0, "text": "general kenobi.", "speaker": "SPEAKER_01"}
+			]
+		}`))
+	}))
+	defer srv.Close()
+
+	p := New(WithBaseURL(srv.URL), WithHTTPClient(srv.Client()))
+	dm, err := p.DiarizationModel("whisperx-diarize")
+	if err != nil {
+		t.Fatalf("DiarizationModel: %v", err)
+	}
+	res, err := dm.Diarize(context.Background(),
+		audio.DiarizationRequest{Audio: []byte("RIFFfake"), MIME: "audio/wav"},
+		audio.WithSpeakerBounds(2, 4), audio.WithDiarizationLanguage("en"))
+	if err != nil {
+		t.Fatalf("Diarize: %v", err)
+	}
+	if gotPath != "/upstream/whisperx-diarize/asr" {
+		t.Errorf("path = %q", gotPath)
+	}
+	for k, want := range map[string]string{
+		"output": "json", "diarize": "true",
+		"min_speakers": "2", "max_speakers": "4", "language": "en",
+	} {
+		if got := gotQuery[k]; len(got) != 1 || got[0] != want {
+			t.Errorf("query %s = %v, want %q", k, got, want)
+		}
+	}
+	if gotField != "audio_file" {
+		t.Error("audio_file part missing")
+	}
+	if len(res.Segments) != 2 || res.Segments[1].Speaker != "SPEAKER_01" {
+		t.Fatalf("segments = %+v", res.Segments)
+	}
+	if res.Language != "en" || res.Text == "" {
+		t.Errorf("language/text = %q/%q", res.Language, res.Text)
+	}
+}
+
+func TestDiarizeRejectsBadArgs(t *testing.T) {
+	p := New(WithBaseURL("http://unused"))
+	dm, _ := p.DiarizationModel("whisperx-diarize")
+	if _, err := dm.Diarize(context.Background(), audio.DiarizationRequest{}); !errors.Is(err, llm.ErrUnsupported) {
+		t.Errorf("no audio: err = %v, want ErrUnsupported", err)
+	}
+	if _, err := dm.Diarize(context.Background(),
+		audio.DiarizationRequest{Audio: []byte{1}, MinSpeakers: 5, MaxSpeakers: 2}); !errors.Is(err, llm.ErrUnsupported) {
+		t.Errorf("bounds 5>2: err = %v, want ErrUnsupported", err)
+	}
+}
+
+func TestDiarizeEmptyTranscriptIsError(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		_, _ = w.Write([]byte(`{"text": "", "segments": []}`))
+	}))
+	defer srv.Close()
+
+	p := New(WithBaseURL(srv.URL), WithHTTPClient(srv.Client()))
+	dm, _ := p.DiarizationModel("whisperx-diarize")
+	_, err := dm.Diarize(context.Background(), audio.DiarizationRequest{Audio: []byte{1}})
+	var apiErr *llm.APIError
+	if !errors.As(err, &apiErr) {
+		t.Fatalf("err = %v, want APIError for empty transcript", err)
+	}
+}