[tanvir-ux]

Summary

Two gaps in the Java Pulse models for piped streams:

1. PipedReader() and PipedWriter() (no-arg constructors) were not tracked as resources. Only the constructors that take the other end were covered, through the generic Reader/Writer delegation model.
2. connect on PipedInputStream, PipedOutputStream, PipedReader and PipedWriter was not modeled, so Pulse did not know it can throw IOException.

Because of this, a pipe end created with the no-arg constructor and closed only on the normal path after connect was never reported. FN_pipedReaderNotConnectedBad and FN_pipedWriterNotClosedAfterConnectBad show this.

• Model the 1-argument <init> (just this) of PipedReader/PipedWriter with Resource.allocate. These constructors don't throw. The constructors that take the other end keep the existing delegation model.
• Model connect on all four piped classes with Resource.use ~exn_class_name:"java.io.IOException", the same way read/write are modeled.

Tests

• FN_pipedReaderNotConnectedBad and FN_pipedWriterNotClosedAfterConnectBad are renamed without the FN_ prefix and are now reported.
• pipedReaderNotClosedAfterConnectOk was a leak (close() is skipped if connect or read throws), so it is renamed pipedReaderNotClosedAfterReadBad and is now reported.
• New tests pipedInputStreamNotClosedAfterConnectBad and pipedOutputStreamNotClosedAfterConnectBad in ResourceLeaks.java cover connect for the byte streams.
• pipedReaderClosedOk, pipedWriterClosedOk and the other Ok cases stay clean. FN_pipedWriterNotConnectedBad and FN_pipedReaderWriterBad are left as they are.

Test plan

• Before (main @ a8f5b17): none of the 5 tests above is reported
• After: PULSE_RESOURCE_LEAK for all 5, allocated by the PipedReader(), PipedWriter(), PipedInputStream() or PipedOutputStream() constructor
• make direct_java_pulse_test: the only issues.exp changes are those 5 lines. (As in [java] Track PipedInputStream and PipedOutputStream as resources #2290, I still get the unrelated lambda-name differences in InvokeDynamic/TransitiveAccess locally with my javac; not committed.)
• ocamlformat 0.29.0 reports no changes for PulseModelsJava.ml