feat(compression): reject empty compression spec

An empty spec list passed to compress() previously returned an
unmodified model silently. Fail early with a clear error instead,
since an empty spec is almost certainly a mistake.

BUG=part of #3256

Author: rkuester

tensorflow/lite/micro/compression/compress.py

@@ -132,6 +132,11 @@ def compress(model_in: ByteString, specs: Iterable[spec.Tensor]) -> bytearray:
   Returns:
     A compressed flatbuffer with DECODE operators inserted.
   """
+  specs = list(specs)
+  if not specs:
+    raise compressor.CompressionError(
+        "Compression spec is empty; no tensors to compress")
+
   model = model_editor.read(model_in)
   compression_results: dict[tuple[int, int], compressor.CompressionResult] = {}
 

tensorflow/lite/micro/compression/compress_test.py

@@ -313,6 +313,11 @@ def test_ancillary_data_format(self):
     self.assertEqual(dcm_bytes[5] & 0x07, 4)  # bitwidth = 4
     self.assertEqual(dcm_bytes[6], 4)  # stride = num unique values
 
+  def test_empty_spec_raises(self):
+    """Empty compression spec is an error, not a silent no-op."""
+    self.assertRaisesRegex(compressor.CompressionError, "empty",
+                           lambda: compress.compress(self.flatbuffer, []))
+
   def test_smaller_bitwidth_raises(self):
     """Specifying LUT compression with too small a bitwidth fails."""
     specs = [