[Refactor] Output Processor Phase 2: separate multimodal output channel (#1601) (#2744)

Signed-off-by: Megha Agarwal <agarwalmegha1308@gmail.com>
Signed-off-by: meghaagr13 <divyanshsinghvi@gmail.com>
Signed-off-by: Taichang Zhou <tzhouam@connect.ust.hk>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: meghaagr13 <divyanshsinghvi@gmail.com>
Co-authored-by: Boao Shi <aoibosh@connect.hku.hk>
Co-authored-by: Taichang Zhou <tzhouam@connect.ust.hk>
Co-authored-by: Cursor <cursoragent@cursor.com>

tests/engine/test_output_modality.py

@@ -81,7 +81,7 @@ def test_multimodal_payload_and_completion_output():
     assert p is not None
     assert "waveform" in p.tensors and torch.equal(p.primary_tensor, data["waveform"])
     assert p.metadata["sample_rate"] == 16000
-    assert not p.is_empty and len(p) == 1
+    assert not p.is_empty and len(p) == 2  # 1 tensor + 1 metadata
 
     # None/empty returns None
     assert MultimodalPayload.from_dict(None) is None
@@ -125,7 +125,7 @@ def test_output_modality_printed_examples(capsys):
     print(f"    tensors keys : {list(p.tensors.keys())}")
     print(f"    primary_tensor: shape={p.primary_tensor.shape}, dtype={p.primary_tensor.dtype}")
     print(f"    metadata      : {p.metadata}")
-    print(f"    is_empty={p.is_empty}, len={len(p)}")
+    print(f"    is_empty={p.is_empty}, len={len(p)}")  # len = tensors + metadata
     print(f"  from_dict(None) -> {MultimodalPayload.from_dict(None)}")
     print(f"  from_dict({{}})   -> {MultimodalPayload.from_dict({})}")