[rusholme]

What and why

Brief consolidate-1025b, task 1. The rapid_docs pages that state a behaviour or a number a merged PR changed are brought to the merged state, at the behaviour cut-off 2026-10-25 02:00:42 UTC (rapid PRs 445 to 456 on rebuild, rapid_docs 122 to 126 on main, none on rapid_systems). Pages found by grep (overlapfields, own field, proper motion, gaia_, naive, PSF-fit, scalefacref, sca_gain, forced sizes, hand references, SCA in the reference scope); a page with nothing superseded is left alone. decisions.md is untouched (append-only), and the migration page carries no statement these PRs superseded, so it is untouched too.

Files

• system/difference.md: the naive subtraction makes no PSF-fit catalog (its image, uncertainty and SExtractor catalogs stay; bits 4 and 5 never set); the reference PSF goes to ZOGY's and SFFT's catalogs; the finder-fill paragraph no longer names a naive catalog; Measured gains the diff-trim pair; Not decided here restated, with the now-unread output_naive_psfcat_* keys.
• system/reference.md: the scope's SCA is the first SCA among the frames; inputs are of the scope's filter, of any SCA (PR 450 relaxed the stage's check).
• system/intake.md: the gaia_proper_motion setting row (shipped false, the flight overlay true) and the Gaia offset paragraph.
• system/datasets.md: GBTDS SCA 2's offset is not proper motion; OpenUniverse's 2.2 is the socsim reference zero point, with ou-gain's 1.018 and ou-zp's failed acceptance (overlay unchanged) and the sca_gain finding; flight's Gaia row under the overlay's propagation; flight's dither row under the loop's overlapfields rule; run 126 as the run that takes the four epochs through the loop alone.
• system/foundation.md: forced rows are objects times window images, with forced-growth's numbers and the table's size (1.02 GB); the naive item in what remains.

The merged PRs read, and where each landed

PR	landed
rapid 445 flight-first	datasets (flight, already in rapid_docs 123; run 118 now named as the first run)
rapid 446 diff-trim code	difference; foundation (what remains)
rapid 448 ou-gain	datasets (OpenUniverse)
rapid 449 self-contained	no page states more than rapid_docs 124 already wrote
rapid 450 ref-overlap code	reference; datasets (flight); loop.md already current from rapid_docs 125
rapid 451 ou-zp	datasets (OpenUniverse)
rapid 452 ref-overlap runbook	datasets (flight, run 126)
rapid 453 gaia-pm code	intake; datasets (flight)
rapid 454 gaia-pm runbook	datasets (flight, GBTDS)
rapid 455 forced-growth	foundation
rapid 456 diff-trim runbook	difference (Measured)
rapid_docs 122, 123, 124, 125, 126	their own pages, already current; 125's note on datasets.md's flight row is done here
rapid_systems	none merged

Measurements

None new; every number is from the PR named beside it.

Improvements seen

• observability.md's profile (six PSF-fit calls including naive's) is a dated measurement and stays; the next profile of a difference job would show four.

Session notes

none

🤖 Generated with Claude Code