feat(tunnel): switch to yamux stream proxying and connection pooling

- Introduce pooled tunnel sessions (TunnelID/DataConnect) on client/server
- Proxy HTTP/HTTPS via raw HTTP over yamux streams; pipe TCP streams directly
- Move UI/stats into internal/shared; refactor CLI tunnel helpers; drop msgpack/hpack legacy

internal/shared/netutil/pipe.go

@@ -0,0 +1,164 @@
+package netutil
+
+import (
+	"context"
+	"io"
+	"sync"
+	"time"
+
+	"drip/internal/shared/pool"
+)
+
+const tcpWaitTimeout = 10 * time.Second
+
+type closeReader interface {
+	CloseRead() error
+}
+
+type closeWriter interface {
+	CloseWrite() error
+}
+
+type readDeadliner interface {
+	SetReadDeadline(t time.Time) error
+}
+
+// Pipe copies bytes bidirectionally between a and b (gost-like),
+// and applies TCP half-close when supported.
+func Pipe(ctx context.Context, a, b io.ReadWriteCloser) error {
+	return PipeWithCallbacksAndBufferSize(ctx, a, b, pool.SizeMedium, nil, nil)
+}
+
+// PipeWithCallbacks is Pipe with optional byte counters for each direction:
+// onAToB is called with bytes copied from a -> b, onBToA for b -> a.
+func PipeWithCallbacks(ctx context.Context, a, b io.ReadWriteCloser, onAToB func(n int64), onBToA func(n int64)) error {
+	return PipeWithCallbacksAndBufferSize(ctx, a, b, pool.SizeMedium, onAToB, onBToA)
+}
+
+// PipeWithBufferSize is Pipe with a custom buffer size.
+func PipeWithBufferSize(ctx context.Context, a, b io.ReadWriteCloser, bufSize int) error {
+	return PipeWithCallbacksAndBufferSize(ctx, a, b, bufSize, nil, nil)
+}
+
+// PipeWithCallbacksAndBufferSize is PipeWithCallbacks with a custom buffer size.
+func PipeWithCallbacksAndBufferSize(ctx context.Context, a, b io.ReadWriteCloser, bufSize int, onAToB func(n int64), onBToA func(n int64)) error {
+	if bufSize <= 0 {
+		bufSize = pool.SizeMedium
+	}
+	if bufSize > pool.SizeLarge {
+		bufSize = pool.SizeLarge
+	}
+
+	var wg sync.WaitGroup
+	wg.Add(2)
+
+	stopCh := make(chan struct{})
+	var closeOnce sync.Once
+	closeAll := func() {
+		closeOnce.Do(func() {
+			close(stopCh)
+			_ = a.Close()
+			_ = b.Close()
+		})
+	}
+
+	errCh := make(chan error, 2)
+
+	if ctx != nil {
+		go func() {
+			select {
+			case <-ctx.Done():
+				closeAll()
+			case <-stopCh:
+			}
+		}()
+	}
+
+	go func() {
+		defer wg.Done()
+		err := pipeBuffer(b, a, bufSize, onAToB, stopCh)
+		if err != nil {
+			errCh <- err
+		}
+		closeAll()
+	}()
+
+	go func() {
+		defer wg.Done()
+		err := pipeBuffer(a, b, bufSize, onBToA, stopCh)
+		if err != nil {
+			errCh <- err
+		}
+		closeAll()
+	}()
+
+	wg.Wait()
+
+	select {
+	case err := <-errCh:
+		return err
+	default:
+		return nil
+	}
+}
+
+func pipeBuffer(dst io.ReadWriteCloser, src io.ReadWriteCloser, bufSize int, onCopied func(n int64), stopCh <-chan struct{}) error {
+	bufPtr := pool.GetBuffer(bufSize)
+	defer pool.PutBuffer(bufPtr)
+
+	buf := (*bufPtr)[:bufSize]
+	_, err := copyBuffer(dst, src, buf, onCopied, stopCh)
+
+	if cr, ok := src.(closeReader); ok {
+		_ = cr.CloseRead()
+	}
+
+	if cw, ok := dst.(closeWriter); ok {
+		if e := cw.CloseWrite(); e != nil {
+			_ = dst.Close()
+		}
+		if rd, ok := dst.(readDeadliner); ok {
+			_ = rd.SetReadDeadline(time.Now().Add(tcpWaitTimeout))
+		}
+	} else {
+		_ = dst.Close()
+		if rd, ok := dst.(readDeadliner); ok {
+			_ = rd.SetReadDeadline(time.Now().Add(tcpWaitTimeout))
+		}
+	}
+
+	return err
+}
+
+func copyBuffer(dst io.Writer, src io.Reader, buf []byte, onCopied func(n int64), stopCh <-chan struct{}) (written int64, err error) {
+	for {
+		select {
+		case <-stopCh:
+			return written, io.EOF
+		default:
+		}
+
+		nr, er := src.Read(buf)
+		if nr > 0 {
+			nw, ew := dst.Write(buf[:nr])
+			if nw > 0 {
+				written += int64(nw)
+				if onCopied != nil {
+					onCopied(int64(nw))
+				}
+			}
+			if ew != nil {
+				return written, ew
+			}
+			if nr != nw {
+				return written, io.ErrShortWrite
+			}
+		}
+		if er != nil {
+			if er == io.EOF {
+				return written, nil
+			}
+			return written, er
+		}
+	}
+}