feat(build): make the SCTP transport an optional dependency

[Thecave3]

Stacked on 67-68-message-id-correlation (#72) — review that first.

The problem

A downstream container image that builds libe3 without the SCTP headers fails:

src/connector/posix_connector.cpp:19:10:
  fatal error: netinet/sctp.h: No such file or directory

posix_connector.cpp included <netinet/sctp.h> unconditionally, so every consumer had to install
libsctp-dev whether or not it ever used SCTP. Images that happened to install it for another
reason built fine; images that did not could not build libe3 at all — for a transport they were
never going to use.

The change

LIBE3_ENABLE_SCTP, default OFF. SCTP is the only libe3 dependency that needs a kernel module
plus a distro -dev package for a single system header, while deployments typically run the E3
link over IPC or TCP.

E3TransportLayer::SCTP stays in the enum, at value 0. It is part of the ABI and of the
wire-facing value the C API (transport_layer: 0=SCTP) and the Python bindings mirror, so removing
or renumbering it would break them. transport_layer_to_string and its test are untouched.

With the flag off, asking for the SCTP transport fails at socket creation with a message naming
the flag — rather than falling through to another protocol, which would look like it worked while
putting traffic on the wrong one. The four socket() sites now go through one sctp_stream_socket()
shim returning -1 with errno=EPROTONOSUPPORT, which every existing caller's error path already
handles.

Worth noting: libe3 calls no libsctp function — only IPPROTO_SCTP and SCTP_NODELAY, both
from the header — so there is nothing to link and no find_package to add. The dependency is the
header alone. It is checked at configure time, so -DLIBE3_ENABLE_SCTP=ON without it fails naming
the package per distro instead of dying partway through the build.

The define is PUBLIC, so it reaches libe3.pc (same pattern as LIBE3_ENABLE_LATREC): a consumer
can tell whether asking for SCTP will work instead of finding out at socket().

build_libe3 gains --enable-sctp / --disable-sctp and no longer installs the SCTP headers
with --install-deps. README.md documents the flag, the package per distro, and that the enum
survives either way.

Verification

Run in a container with no libsctp-dev — the environment that was failing:

arm	result
default (SCTP off)	configures and builds to completion
-DLIBE3_ENABLE_SCTP=ON	fails at configure: "LIBE3_ENABLE_SCTP=ON but <netinet/sctp.h> was not found. Install the SCTP headers (Debian/Ubuntu: libsctp-dev, …)"

Also: preprocessing posix_connector.cpp with the flag off shows netinet/sctp.h is never reached,
and both arms build on a host that does have the header.

Consumer impact

None required. A consumer that already passes explicit -DLIBE3_* flags simply gets the
default-off behaviour; one that genuinely needs SCTP adds -DLIBE3_ENABLE_SCTP=ON and the distro
package.

[github-actions]

CI report — 44d1684 — ✅ all checks passed

Workflow	Result	Time	Run
Commit policy	✅ success	3m36s	#95
E2E dApp Integration	✅ success	1m57s	#110
E2E Topologies (multi-dApp / multi-RAN)	✅ success	2m26s	#108
Full-loop Latency Benchmark	✅ success	1m13s	#109
Unit Tests	✅ success	4m40s	#136
latrec portability	⏭️ not triggered (paths filter)	—	—
MPMC Queue Benchmark	⏭️ not triggered (paths filter)	—	—

E2E Topologies (multi-dApp / multi-RAN)

zmq/ipc

• ✅ 1 RAN - 1 dApp: indications=5
  • dapp peer=t11 ran=ran-solo sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=1 max=1 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]
• ✅ 1 RAN - 2 dApps: dApp#1 ind=5 sub=1, dApp#2 ind=6 sub=2, RAN saw 2 dApps
  • dapp1 peer=t12 ran=ran-shared sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0 max=0 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]
  • dapp2 peer=t12 ran=ran-shared sub=2 indications=6 seq=[0..5] dropped=0 (0%) age_ms(avg=0 max=0 @seq=0) hist[<=1:6 2-5:0 6-10:0 >10:0]
• ✅ 2 RANs - 1 dApp: from ran-a ind=5, from ran-b ind=5
  • dapp peer=t2a ran=ran-a sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0 max=0 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]
  • dapp peer=t2b ran=ran-b sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0 max=0 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]

zmq/tcp

• ✅ 1 RAN - 1 dApp: indications=5
  • dapp peer=default ran=ran-solo sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0.8 max=1 @seq=1) hist[<=1:5 2-5:0 6-10:0 >10:0]
• ✅ 1 RAN - 2 dApps: dApp#1 ind=5 sub=1, dApp#2 ind=6 sub=2, RAN saw 2 dApps
  • dapp1 peer=default ran=ran-shared sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0.2 max=1 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]
  • dapp2 peer=default ran=ran-shared sub=2 indications=6 seq=[0..5] dropped=0 (0%) age_ms(avg=0.166667 max=1 @seq=0) hist[<=1:6 2-5:0 6-10:0 >10:0]
• ✅ 2 RANs - 1 dApp: from ran-a ind=5, from ran-b ind=5
  • dapp peer=default ran=ran-a sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0.4 max=1 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]
  • dapp peer=off100 ran=ran-b sub=1 indications=5 seq=[0..4] dropped=0 (0%) age_ms(avg=0 max=0 @seq=0) hist[<=1:5 2-5:0 6-10:0 >10:0]

E2E dApp Integration

✅ posix/ipc

• dApp exit: 0
• Indications received: 7

✅ posix/tcp

• dApp exit: 0
• Indications received: 7

✅ zmq/ipc

• dApp exit: 0
• Indications received: 7

✅ zmq/tcp

• dApp exit: 0
• Indications received: 7

Full-loop Latency Benchmark

Full-loop latency

Full-loop latency benchmark (N=1014 after 50 warmup)

All values in microseconds (μs). Link: zmq, transport: ipc, encoding: ASN.1 APER.

#	Description	Tags	mean	p50	p99	max
1	Collect indication data	RECORD_BEGIN to ENCODE_E3SM_BEGIN	0.13	0.11	0.35	2.36
2	Create & encode indication	ENCODE_E3SM_BEGIN to ENCODE_E3SM_DONE	1.05	0.98	1.73	33.72
3	Encode E3AP (indication)	EMIT_ENTER to ENQUEUE, then DEQUEUE to ENCODE_E3AP_DONE	3.11	3.03	5.78	6.23
4	Queuing (indication)	ENQUEUE to DEQUEUE	5.18	4.48	14.23	62.68
5	Delivery (indication)	ENCODE_E3AP_DONE to SEND_DONE	0.34	0.28	1.70	7.09
6	E3 wire (RAN -> dApp)	SEND_DONE to RECV	52.56	52.75	65.54	84.70
7	Decode E3AP (indication)	RECV to DECODE_E3AP_DONE	1.79	1.50	2.96	22.74
8	libe3 dispatch (indication)	DECODE_E3AP_DONE to DELIVER_BEGIN	0.13	0.13	0.22	0.47
9	Decode indication	DELIVER_BEGIN to DECODE_E3SM_DONE	0.67	0.60	1.05	1.41
10	Process data	DECODE_E3SM_DONE to ENCODE_E3SM_BEGIN	0.04	0.04	0.05	0.13
11	Create & encode control	ENCODE_E3SM_BEGIN to ENCODE_E3SM_DONE	0.34	0.31	0.60	0.78
12	Encode E3AP (control)	EMIT_ENTER to ENQUEUE, then DEQUEUE to ENCODE_E3AP_DONE	3.79	3.77	5.43	6.72
13	Queuing (control)	ENQUEUE to DEQUEUE	16.66	16.50	23.37	67.97
14	Delivery (control)	ENCODE_E3AP_DONE to SEND_DONE	5.01	4.86	8.12	9.82
15	E3 wire (dApp -> RAN)	SEND_DONE to RECV	51.47	50.82	65.22	98.74
16	Decode E3AP (control)	RECV to DECODE_E3AP_DONE	2.76	2.60	4.95	15.54
17	libe3 dispatch (control)	DECODE_E3AP_DONE to DECODE_E3SM_BEGIN	0.27	0.25	0.52	0.64
18	Decode & handle control	DECODE_E3SM_BEGIN to DECODE_E3SM_DONE	0.42	0.38	0.70	0.91
Total	Total round-trip		146.23	147.34	181.33	224.11

ubuntu-latest, Release build, ZMQ + IPC, ASN.1 APER.

These numbers are measured inside a GitHub Actions container and should be treated as an upper bound on E3AP's and the library's own latency, not a representative deployment measurement.

Ready to merge (fast-forward only)

A maintainer can land the reviewed commits with:

git fetch origin
git checkout main && git merge --ff-only 44d168446942dc4a2fa9ecac938ec38cbe24b3cc && git push origin main

Head: 44d168446942dc4a2fa9ecac938ec38cbe24b3cc (branch feat/optional-sctp). If --ff-only fails as non-fast-forward, the branch must be rebased on the latest main.

One comment per PR, rewritten in place once every workflow for 44d1684 finished.