test(server): bind HTTP fixtures to IPv4 loopback

[rynfar]

NodeHttpServer.layerTest can bind a test server to IPv6 :: while its Fetch client requests 127.0.0.1. On macOS, another process can bind the same numeric port on IPv4, so a server seam test can receive that other listener's body. The full server.test.ts suite intermittently received ready or an unrelated 501 response; a same-process two-listener reproduction confirmed the address-family collision.

This test-only change binds server fixtures to 127.0.0.1, matching the client, across the server seam, analytics, and MCP HTTP tests. The regression reads the actual server address and verifies that a competing IPv4 bind fails with EADDRINUSE. It leaves production HTTP behavior unchanged and adds no retries or timing delays. The fixture is compatible with Effect rc.112 and rc.115 address types; on the reviewed open-PR integration stack, an additional provider-refresh seam test must use the same helper when those branches converge.

Prerequisite: #743 adds a provider-refresh seam test. This branch merges its exact reviewed head and switches that added test to the same loopback layer, so integration does not leave a dangling NodeHttpServer.layerTest reference. The PR still targets pylon; #743 must land before this dependent patch is merged.

Validation: 240 focused tests and t3 typecheck passed on the combined #743 + fix branch; scoped lint, format, and frozen install passed. Six bounded 201-test full-file runs passed after the initial change. An isolated rc.115 integration rehearsal passed 223 focused tests and t3 typecheck. Independent review passed on the final combined commit. This reduces a confirmed test-fixture collision; repeated passing tests alone do not establish that every intermittent failure is eliminated.

Tracks upstream catch-up issue #736 only as a test-isolation follow-up; no upstream source disposition or ledger cursor is changed.

[vercel]

Deployment failed for project pylon-marketing with the following error:

Resource is limited - try again in 24 hours (more than 100, code: "api-deployments-free-per-day").

Learn More: https://vercel.com/pylon-code?upgradeToPro=build-rate-limit

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	14.0 KiB	13.9 KiB	−40 B (−0.3%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.2 KiB	7.2 KiB	0 B (0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.8 KiB	6.7 KiB	−40 B (−0.6%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	58.0 KiB	58.0 KiB	−44 B (−0.1%)	66.4 KiB	✅
Codex	Live turn messages	10	9	−1 (−10.0%)	21	✅
Claude	Total thread wire	14.0 KiB	14.0 KiB	−16 B (−0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.2 KiB	7.2 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.7 KiB	6.7 KiB	−8 B (−0.1%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	58.9 KiB	58.9 KiB	−44 B (−0.1%)	66.4 KiB	✅
Claude	Live turn messages	10	9	−1 (−10.0%)	21	✅

Baseline: 7ed3f71 · PR result: cd2441f · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 115.6 KiB
• Claude decoded thread snapshot: 116.3 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.