Use REPLACE for malformed bytes so truncation preserves declared charset

When FileItemContentReader truncates at MAX_CONTENT_BYTES a cut in the
middle of a multibyte character no longer triggers the fallback path.
REPLACE substitutes the incomplete sequence with U+FFFD using the
declared charset; REPORT was throwing and silently switching to the
JVM default charset for the whole string.

Author: jandro996

internal-api/src/main/java/datadog/trace/api/http/MultipartContentDecoder.java

@@ -14,8 +14,8 @@ public static String decodeBytes(byte[] buf, int length, String contentType) {
     try {
       return charset
           .newDecoder()
-          .onMalformedInput(CodingErrorAction.REPORT)
-          .onUnmappableCharacter(CodingErrorAction.REPORT)
+          .onMalformedInput(CodingErrorAction.REPLACE)
+          .onUnmappableCharacter(CodingErrorAction.REPLACE)
           .decode(ByteBuffer.wrap(buf, 0, length))
           .toString();
     } catch (CharacterCodingException e) {

internal-api/src/test/java/datadog/trace/api/http/MultipartContentDecoderTest.java

@@ -64,14 +64,24 @@ void decodeBytesUsesMachineDefaultFallbackWhenBytesCannotBeDecoded() {
   }
 
   @Test
-  void decodeBytesUsesMachineDefaultFallbackWhenDeclaredCharsetCannotDecodeBytes() {
+  void decodeBytesReplacesMalformedBytesWithReplacementCharacterUsingDeclaredCharset() {
+    // 0xE9 (ISO-8859-1 'é') is not valid UTF-8; REPLACE substitutes U+FFFD
     byte[] bytes = "café".getBytes(StandardCharsets.ISO_8859_1);
-    String expected = new String(bytes, Charset.defaultCharset());
     assertEquals(
-        expected,
+        "caf�",
         MultipartContentDecoder.decodeBytes(bytes, bytes.length, "text/plain; charset=UTF-8"));
   }
 
+  @Test
+  void decodeBytesHandlesTruncationAtMultibyteCharacterBoundary() {
+    // "€" encodes as 3 bytes in UTF-8: E2 82 AC
+    byte[] complete = "hello€".getBytes(StandardCharsets.UTF_8); // 8 bytes
+    // Pass only 6 bytes: "hello" + first byte of "€" (incomplete sequence)
+    String result = MultipartContentDecoder.decodeBytes(complete, 6, "text/plain; charset=UTF-8");
+    // Incomplete sequence → U+FFFD with declared charset, not fallback to JVM default
+    assertEquals("hello�", result);
+  }
+
   @ParameterizedTest
   @NullAndEmptySource
   void extractCharsetReturnsNullForNullOrEmptyContentType(String contentType) {