Fix incomplete data detection to handle all decoder exceptions

Extended the incomplete data detection logic to catch ALL types of insufficient data errors from the decoder, not just "not long enough" messages.

The decoder throws different incomplete data errors:
- "Content not long enough" - when parsing message header (13 bytes needed)
- "Segment header is incomplete" - when parsing segment header (10 bytes needed)
- "Segment footer is incomplete" - when parsing segment footer (8 bytes for CRC64)
- "Message footer is incomplete" - when parsing message footer (8 bytes for CRC64)

Previous fix only caught "not long enough", causing failures when interruptions occurred while parsing segments/footers.

Updated exception handling to check for both:
- "not long enough" (header parsing)
- "is incomplete" (segment/footer parsing)

This ensures partial data is preserved in pending buffer for ALL incomplete data scenarios, allowing smart retry to properly accumulate bytes across interruptions regardless of where in the message structure the interruption occurs.

Addresses the "Unexpected segment number" error that occurred when segments were split across retries.

Co-authored-by: gunjansingh-msft <[email protected]>

Author: Copilot

sdk/storage/azure-storage-common/src/main/java/com/azure/storage/common/policy/StorageContentValidationDecoderPolicy.java

@@ -116,7 +116,8 @@ private Flux<ByteBuffer> decodeStream(Flux<ByteBuffer> encodedFlux, DecoderState
                 }
             } catch (IllegalArgumentException e) {
                 // Handle decoder exceptions - check if it's due to incomplete data
-                if (e.getMessage() != null && e.getMessage().contains("not long enough")) {
+                String errorMsg = e.getMessage();
+                if (errorMsg != null && (errorMsg.contains("not long enough") || errorMsg.contains("is incomplete"))) {
                     // Not enough data to decode yet - preserve all data in pending buffer
                     state.updatePendingBuffer(dataToProcess);