docs(fep): QuantPub rev 6 — pre-submission consistency fixes (#905)

[fiddur]

Part of #905 (does not close it). Batches the non-blocking items from the #1038 review, per the reviewer's note to fold them in before upstream submission:

1. Motivation ↔ Implementations self-contradiction fixed: the Motivation section no longer claims Aurboda ships a "vendor-prefixed (aurboda:) variant" — the quant: vocabulary is on the wire since feat(feed): federate the QuantPub quant:Exercise extension + serve its @context (#896) #1034. Same stale wording fixed in the timeline-enrich.ts header (also the first checkbox of QuantPub extension polish (deferred review nits from #1034) #1037).
2. docs/fep/README.md now indexes quantpub-reddit-draft.md and states it is not itself a FEP.
3. Reddit-draft overstatements: "static routes" → "routes" (the series endpoint carries the §6 authorization check); "no-store everywhere" → no-store on data responses with the cacheable discovery doc called out; workouts-are-running-code vs any-observation-is-spec claims separated.
4. Item 5 (consumer-side quant:structuredUrl token fallback) recorded on QuantPub extension polish (deferred review nits from #1034) #1037 as a code follow-up rather than changed here.

Backend pnpm check green (comment-only code change).

🤖 Generated with Claude Code

https://claude.ai/code/session_016VFPcqXTu9DPuARhRBnBoo

[fiddur]

✅ Approved

Docs-only change plus one doc-comment rewrite; no behavior changes. I verified each claim the diff makes against the code:

• timeline-enrich.ts: the new header is accurate — quant-extension.ts does put the scalar summaries in-band (quant:metrics) while quant:series carries only links, so "scalar summary in-band, not the high-resolution series" is right, and the fetch path (api_base + /public/{user}/feed/{postId}, matched by FEED_OBJECT_PATH) is indeed the §7 id-convention shape. The old text ("Fedify's typed vocab drops the aurboda: extension") was genuinely stale after #896, so this is a real fix rather than comment churn.
• quantpub-reddit-draft.md: no-store on every data response / discovery-only caching matches §8.4 and §4; dropping "static" from "two-and-a-half static routes" matches the weekend section (only the discovery doc is static); and narrowing "my workouts and sleep observations are ActivityPub posts" to workouts matches what Aurboda actually federates today.
• quantpub.md: the quant: wire-vocabulary claim matches the Implementations section and quant-extension.ts.
• docs/fep/README.md: the new entry correctly labels the draft as not-a-FEP.

Nothing blocking. Approving.

Non-blocking — fold into a later PR, don't re-roll this one

• docs/fep/quantpub-reddit-draft.md:17: the rewritten sentence leaves one ~110-col line in a file otherwise wrapped at ~80. Also, the new parenthetical "(and the spec covers any observation — sleep, HRV, steps, mood, …)" now largely duplicates the later privacy-list bullet "It's not just exercise: a generic Observation shape covers sleep, HRV, steps, mood, glucose"; one of the two could go.
• docs/fep/quantpub.md Implementations (outside this diff): the parenthetical says the remaining vendor surface is "the /.well-known/aurboda document backs Aurboda-specific challenge federation", but the Level 3 enrichment path also uses it — realEnrichDeps wires discover: discoverInstance, which fetches /.well-known/aurboda (challenge-federation.ts:72), not §4's /.well-known/quantpub. So the shipped ingest resolves api_base from the vendor document rather than the spec one. Since the PR's theme is pre-submission accuracy, that's worth reconciling before FEP submission — either point discoverInstance/enrichment at /.well-known/quantpub (it is already served, well-known-router.ts:73) or soften the Implementations wording.