trickle: Add a /<channel>/next endpoint

[rickstaa]

Extracted from #3884 — a small, independent piece that unblocks the livepeer-python-gateway pipeline SDK without waiting for the rest of the "Serverless" work.

What this adds

GET /{streamName}/next returns the channel's current nextWrite:

• Body: the next seq as plain text
• Header: Lp-Trickle-Latest: {seq}
• 404 if the channel doesn't exist
• Lp-Trickle-Closed: terminated if the channel is closed

Why now

The Python SDK (livepeer-python-gateway) probes /next to learn nextWrite before its first POST so it doesn't collide with the server's own -1-resolves-to-slot-0 behavior. Without this endpoint on master, the SDK's probe fails (the path matches the {idx} route, fails to parse "next" as an int, returns 400) and the publisher's fallback drops the first segment on every channel.

Companion SDK fallback fix: livepeer/livepeer-python-gateway#14
Bug context: livepeer/livepeer-python-gateway#12

Relationship to #3884

Cherry-picked from commit 8da1c69224a866386770927de30f43bb565508cf on the ja/serverless branch. The Lp-Trickle-Reset hunk in handlePost that travelled with that commit on the branch belongs to a separate "trickle: Support publisher resets" feature and was excluded here — landing it would expand scope beyond what the SDK needs.

When #3884 lands, this commit will already be on master; the rebase should be clean.

Co-author

Original code by @j0sh — preserved as commit author. Extraction + adapted commit message by Rick Staa, with this PR opened on Josh's behalf to unblock dependent SDK work.

Test plan

• go build ./trickle/... passes
• go test ./trickle/... passes
• Manual: build a :trickle-next-endpoint orchestrator image, point the SDK at it, confirm live_transcribe delivers transcripts 0–3 (vs. the current 0, 2, 3 due to the slot-0 race)

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 6.66667% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 32.88179%. Comparing base (9e68815) to head (5d82a42).

Files with missing lines	Patch %	Lines
trickle/trickle_server.go	6.66667%	14 Missing ⚠️

Additional details and impacted files

@@                 Coverage Diff                 @@
##              master       #3925         +/-   ##
===================================================
+ Coverage   32.87450%   32.88179%   +0.00729%     
===================================================
  Files            171         171                 
  Lines          42063       42078         +15     
===================================================
+ Hits           13828       13836          +8     
- Misses         27194       27199          +5     
- Partials        1041        1043          +2     

Files with missing lines	Coverage Δ
trickle/trickle_server.go	69.61039% <6.66667%> (-2.55177%)	⬇️

... and 6 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9e68815...5d82a42. Read the comment docs.

Files with missing lines	Coverage Δ
trickle/trickle_server.go	69.61039% <6.66667%> (-2.55177%)	⬇️

... and 6 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[rickstaa]

@j0sh quick fix to make livepeer-python-gateway PR #7 compatible with the go-livepeer side. I didn’t aim for full code parity yet, but it might make sense to move both the Python and Go trickle subscribers into a separate shared package so they don’t diverge over time. Curious to hear what you think.