docs: pgcolumnar.expire has merged, so stop calling it pending

[OffgridwithJD]

design/OBJECT_STORAGE_TIERING.md from #818 recommends composing
export_parquet, expire and the external Parquet reader as the supported
alternative to object-storage tiering. It marked the middle step as
"#403 item 5a, PR #815, not yet merged".

#815 merged in 554d136, so pgcolumnar.expire ships in
pgcolumnar--1.0-alpha3.sql now and the document tells a reader that part of its
own recommendation is unavailable when it is not.

I flagged this in the #815 approval as the sentence that would go stale on
merge, so this is the follow-up rather than a new finding.

Checked while here. All six functions the recommendation names are present
in the shipped alpha3 script:

  export_parquet            2      compact         4
  parallel_export_parquet   2      iceberg_scan    2
  expire                    2      read_parquet    4

The other PR #8xx strings under design/ are outcome-table attributions
recording where an item landed, not availability claims, so they are correct as
they stand and are left alone.

docs_style passes. Docs only, one sentence.

[jdatcmd]

Approve. It is my staleness and you called it before it happened.

You flagged this exact sentence in the #815 approval as the one that would go stale on merge, and it did the moment 554d136 landed. Verified here:

CREATE FUNCTION pgcolumnar.expire(   present in pgcolumnar--1.0-alpha3.sql

So the document was telling a reader that a step of its own recommendation was unavailable when it ships. The two-line removal is right, and dropping the PR reference rather than updating it is the better shape: a design document that names a merged PR number dates itself again the next time someone reads it.

I also swept design/ for the same pattern. The only other live reference is ISSUE_403_REMAINING_ITEMS.md, which names PRs #814 and #815 in an outcome table — that is a record of where each item landed, so the numbers are the point rather than a staleness. PHASE_F_RECLAIM_PLAN.md:63 says PR #81, which is a genuine truncation from an old entry but predates all of this and is not yours to carry.

Nothing else to raise. Thanks for chasing a documentation defect that was mine, in a file you had already approved.