datapath: fix 12 correctness findings from tun/UDP offload review

Multi-disciplinary correctness review of the batched tun / GSO-GRO / sendmmsg
rework. Each fix has a regression test; the merged tree builds on
linux/darwin/openbsd/windows/freebsd/netbsd, vets clean, passes the unit and
e2e suites, and is -race clean.

Critical:
- C1 zero-length inner UDP datagram no longer panics the process (remote DoS):
  the UDP coalescer routes payLen==0 to passthrough instead of seeding a GSO
  slot, and WriteGSO skips empty payload iovecs as defense in depth.
- C2 segmenter no longer corrupts inner headers when gsoSize < headerLen: the
  L3+L4 header is snapshotted once and each segment stamped from the copy,
  replacing the destructive overlapping in-place slide (SegmentTCP + SegmentUDP).

High:
- H1 applyOuterECN updates the IPv4 header checksum (RFC 1624 incremental) when
  folding outer CE into the inner ToS, so passthrough packets are no longer
  dropped by the peer stack.
- H2 the GRO reject path caps the borrowed RX segment ([:n:n]) so a reject can
  no longer overrun into the next coalesced segment's Nebula header. Note:
  oversized ICMPv6 rejects that need >16B beyond the segment are now refused
  rather than sent under GRO (safe; see TOFIX.md for the scratch-buffer follow-up).
- H3 WriteBatch falls back to per-packet WriteTo for a chunk when writeSockaddr
  fails, so one bad-family destination costs only its own packet, not the batch.
- H4 UserDevice.Readers returns N distinct queue wrappers with private buffers
  (sharing the pipes) so concurrent readers no longer race/overwrite borrowed
  packet bytes.
- H5 Poll.Close / Offload.Close no longer null t.fd (matching master's
  tunFile.Close), removing the data race with a concurrent readOne load.

Medium/Low:
- M1 the UDP GSO 127-segment gate moved from kernel >=5.5 to >=6.9 (the real
  UDP_MAX_SEGMENTS 64->128 threshold), avoiding EINVAL + per-packet fallback on
  5.5-6.8 kernels.
- M2 NewMultiQueueReader replays the offload mask newTun actually negotiated
  instead of the TSO-only mask, so adding a queue no longer disables USO
  device-wide; the advertised USO capability derives from the same mask.
- M3 the shutdown eventfd is closed in pollQueueSet.Close / offloadQueueSet.Close
  (double-close guarded), fixing the per-lifecycle fd leak.
- M4 dual-stack ECN selects the cmsg by address family, not socket family: RX
  parseRecvCmsg reads both IP_TOS and IPV6_TCLASS; TX writeEntryCmsg stamps
  IP_TOS for v4/v4-mapped dests and IPV6_TCLASS for v6 (on-host verified).
- L1 newPoll no longer closes the fd on failure (matching newOffload), removing
  the double-close on QueueSet.Add error.

iputil/packet_test.go

@@ -1,6 +1,7 @@
 package iputil
 
 import (
+	"bytes"
 	"encoding/binary"
 	"net"
 	"testing"
@@ -179,6 +180,51 @@ func Test_CreateRejectPacket_NoICMPError(t *testing.T) {
 	}
 }
 
+// Test_CreateRejectPacket_RespectsCap guards against H2: with UDP GRO the
+// scratch buffer reused to build a reject is a single coalesced segment inside
+// a shared recvmmsg row. Its length covers just that segment, but an uncapped
+// slice's capacity runs on into the next, not-yet-processed segment. Because
+// CreateRejectPacket honors cap, capping the borrowed segment to its own length
+// (cap==len) makes it physically impossible for an oversized ICMPv6 reject to
+// overwrite the neighbor segment's bytes.
+func Test_CreateRejectPacket_RespectsCap(t *testing.T) {
+	src := net.ParseIP("fd00::1")
+	dst := net.ParseIP("fd00::2")
+
+	// Inner IPv6 UDP packet. An ICMPv6 reject copies the whole inner packet
+	// plus a 48-byte header (40 IPv6 + 8 ICMPv6), so it needs 48 more bytes
+	// than the inner packet length.
+	inner := makeIPv6Packet(src, dst, 17, make([]byte, 20))
+
+	// The ciphertext scratch reused as the reject buffer is the received
+	// datagram: 16-byte Nebula header + inner + 16-byte AEAD tag. That is only
+	// 32 bytes of slack, so a full ICMPv6 reject overruns it by 16 bytes.
+	const nebulaOverhead = 32
+	segLen := len(inner) + nebulaOverhead
+
+	// Shared backing row laid out as [segment][neighbor's 16-byte Nebula header].
+	const neighborHdr = 16
+	sentinel := bytes.Repeat([]byte{0xAB}, neighborHdr)
+
+	// Uncapped: the slice's capacity reaches into the neighbor, reproducing
+	// the overrun that silently drops the neighbor packet.
+	backing := make([]byte, segLen+neighborHdr)
+	copy(backing[segLen:], sentinel)
+	reject := CreateRejectPacket(inner, backing[:segLen])
+	assert.NotNil(t, reject, "uncapped buffer reaches into the neighbor, so the reject is built")
+	assert.NotEqual(t, sentinel, backing[segLen:segLen+neighborHdr],
+		"without the cap the oversized reject overruns into the neighbor segment")
+
+	// Capped (the fix): cap==len, so the builder cannot exceed the segment. The
+	// reject does not fit, so it is refused rather than corrupting the neighbor.
+	backing = make([]byte, segLen+neighborHdr)
+	copy(backing[segLen:], sentinel)
+	reject = CreateRejectPacket(inner, backing[:segLen:segLen])
+	assert.Nil(t, reject, "capped segment is 16 bytes too small for a full ICMPv6 reject, so it is refused")
+	assert.Equal(t, sentinel, backing[segLen:segLen+neighborHdr],
+		"capped segment must leave the neighbor untouched")
+}
+
 func makeIPv6Packet(src, dst net.IP, nextHeader uint8, payload []byte) []byte {
 	b := make([]byte, ipv6.HeaderLen+len(payload))
 	b[0] = ipv6.Version << 4