feat(cli): add gzip compression to v2 docs publish requests

[broady]

Description

Refs FER-10484

Adds transparent gzip compression to the v2 (current default) docs publish request bodies. This is the companion to #16033 which does the same for the docs-ledger write path.

Changes Made

• New compressedFetch.ts: createGzipFetch() returns a fetch wrapper that pipes request bodies through CompressionStream("gzip") and sets Content-Encoding: gzip. Also exports gzipJsonBody() for raw fetch calls.
• publishDocs.ts: The FdrClient is constructed while globalThis.fetch is temporarily replaced with the gzip fetch, so oRPC's LinkFetchClient captures it for all subsequent calls (register, finish, API registration).
• publishDocs.ts: The raw fetch translation registration call now uses gzipJsonBody() to compress its JSON body.
• Changelog entry added.

How it works

1. oRPC's LinkFetchClient snapshots globalThis.fetch at construction time
2. We temporarily swap in a gzip-compressing fetch before building the FdrClient, then restore immediately after
3. All subsequent requests through the oRPC client use the captured gzip fetch — bodies are piped through CompressionStream("gzip") with Content-Encoding: gzip header
4. FDR's @fastify/compress transparently decompresses on the server side
5. File uploads to S3 (via axios) and dynamic IR uploads are unaffected

Testing

• Biome lint, format, and check all pass
• Manual testing completed

Link to Devin session: https://app.devin.ai/sessions/08fa84ed296b46629c0316c86ee1aca1
Requested by: @broady

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[claude]

Claude Code Review

This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.

_{Tip: disable this comment in your organization's Code Review settings.}

[devin-ai-integration]

Test Results: Gzip Compression for V2 Docs Publish

Tested by building the dev CLI with these changes and running end-to-end docs publish against registry-dev2.buildwithfern.com using DEV_SMOKE_TEST_FERN_TOKEN.

Content-Length Investigation (Key Concern)

Finding: Content-Length was never set in the first place — the "loss" concern is moot.

• Node.js Request objects do NOT auto-set Content-Length for string bodies
• oRPC's toFetchBody only sets Content-Length for Blob bodies, not JSON
• Our headers.delete("Content-Length") is purely defensive (deletes a header that was never there)
• Node.js fetch automatically uses transfer-encoding: chunked for streaming bodies

Test 1: Unit-Level Transport Verification

Local HTTP test server receiving gzip-compressed requests via the exact createGzipFetch() pattern:

• Content-Encoding: gzip — present ✓
• Content-Length — NOT PRESENT (never was) ✓
• transfer-encoding: chunked — present (auto-set by Node.js) ✓
• Decompressed body matches original payload exactly (1027 bytes) ✓

Test 2: E2E Docs Publish — Minimal Fixture (1 page)

2 FDR POST requests, all gzip'd, all 200 OK:

POST /v2/registry/docs/preview/init  | Content-Encoding: gzip | Content-Length: NOT SET | 200 OK
POST /v2/registry/docs/register/{id} | Content-Encoding: gzip | Content-Length: NOT SET | 200 OK

Published to: https://smoke-test-preview-2f0ca998-0a51-4059-8080-afec8491713e.docs.dev.buildwithfern.com

Test 3: E2E Docs Publish — Full Smoke-Test Fixture (multiple APIs, products, pages)

7 FDR POST requests, all gzip'd, all 200 OK:

POST /v2/registry/docs/preview/init   | Content-Encoding: gzip | Content-Length: NOT SET | 200 OK
POST /registry/api/register (x5)      | Content-Encoding: gzip | Content-Length: NOT SET | 200 OK
POST /v2/registry/docs/register/{id}  | Content-Encoding: gzip | Content-Length: NOT SET | 200 OK

14/14 assertions passed. Non-FDR requests (health, slugs, versions) correctly NOT compressed.

Devin session