docs(blog): rename the ~16k measurement's host noun in context-window-is-the-constraint

[os-steve]

Fixes#9305

What

content/blog/context-window-is-the-constraint.mdx is a fourth host of the
~16k-token examples/app-crm measurement (README.md, metadata-driven.mdx,
how-ai-development-works.mdx are the other three, all already renamed by
#9293). The blog post still called the measured app "the entire business
system", which collides with the adopted "a complete CRM in under 150k
tokens" tagline from #9242 -- two claims both asserting to be the whole
business system, at two different token counts.

Renamed the host noun in the one sentence immediately following the table
(disposition C from the triage comment): "entire business system" ->
"whole example app", matching the exact phrase the other three hosts now use
post-#9293. Nothing else in the post changed -- numbers (31 / 1,792 /
~16,000 / 8% / 200k), the find … wc -l snippet, and the rest of the
argument are untouched. Frontmatter date: 2026-07-17 is untouched; this is
a dated post, not a living doc, so only the characterisation collision is in
scope (not a rewrite of the post's argument).

-That's the **entire business system**. Not a module of it — the data model, the+That's the **whole example app**. Not a module of it — the data model, the
UI, the automation, the access rules, all of it. It fits in about 8% of a
200k-token context window.

Consistency check across all four hosts (read, not swept)

Read how the other three now describe the measurement so this page would not
become a fourth different description:

• README.md: "...is 31 files, 1,792 lines, roughly 16k tokens. That's
  the whole example app, in about 8% of a 200k-token context window."
• content/docs/concepts/metadata-driven.mdx: "That's the whole example
  app -- about 8% of a 200k-token context window."
• content/docs/getting-started/how-ai-development-works.mdx: "...1,792
  lines across 31 files, about 16k tokens. That's the whole example app,
  not a module of it."

All three converge on "the whole example app" as the host-noun phrase; this
PR brings the blog post's leading sentence into the same wording. The four
hosts do not fully diverge otherwise -- no further reconciliation performed
or needed.

Tests

• node scripts/check-nul-bytes.mjs -- OK (6116 tracked text files scanned,
  no raw control bytes) at commit 5c517cf57.
• node scripts/pm/dispatch-gates.mjs content/blog/context-window-is-the-constraint.mdx
  -- no check family names this path directly; content/blog/** is not an
  auto-gen or otherwise gated docs family (content/docs/releases/,
  content/docs/references/, **/translations/*.generated.ts do not apply
  here).
• Diff is a single one-word phrase swap in prose; no package build/typecheck
  applies (docs-only, no code/schema touched).

Generated by Claude Code

---

Generated by Claude Code

[github-actions]

⛔ merge queue 构建失败 — 先分诊，再决定要不要重排

队列构建 32086062275 红了。队列跑的是全量套件（PR 侧 CI 只跑 affected 子集），
所以失败的测试可能在本 PR 没碰过的包里 —— 那不是重排能修的。每次盲目重排都会让排在后面的所有 PR 重建一轮。

失败的 job（日志抽取，best effort）：

• Test Core (2/3) — 失败步骤: Run this shard's tests

  �[41m�[1m FAIL �[22m�[49m src/lint-startup-registry-verdict.corpus.test.ts�[2m > �[22mstartup open-vocabulary verdicts across packages/ (#4776)�[2m > �[22mno package records a verdict the boot can sti
  

历史信号：

• 本 PR 过去 24h 无队列失败记录（首次）。
• 过去 24h 队列共有 5 个失败构建（不含本次）。

分诊清单：

1. 失败测试在本 PR 改动的包里 → 真回归，修 PR。
2. 失败测试与本 PR 无关 → 在其他 PR 的同类评论里搜同名测试；出现过 ⇒ flaky 实锤，开 issue 修/隔离那条测试。修好前重排只会再烧一轮全队列。
3. 两者都不是 → 可能与同组 PR 语义冲突；等前面的 PR 落地或失败出队后再重排一次即可，不要连续重排。

---

Generated by Claude Code · merge-queue-triage workflow (#4859)