Stop calling WASP-108 b imaged, and stop saying no map exists of an imaged exoplanet

48fd6fd took the archive's ima_flag as it stood, so WASP-108 b's card said it had been imaged as a
point of light beside its star. It is a transiting hot Jupiter (period 2.676 d, 0.04 AU out, 0.15 mas
at 258.8 pc). Its flag comes from Bohn et al. 2020 (A&A 635, A73), a VLT/SPHERE survey of transiting
planets' host stars, which imaged a 0.35 solar-mass companion 0.124" from WASP-108, not the planet.
The imaged query now also asks for tran_flag=0. The archive flags exactly one transiting planet as
imaged (queried today: WASP-108 b); with it left out the list has 101 names, the old 102 less that
one, and still above the validator's floor of 95. The new URL is cached under its own name, and
exoplanets.json, regenerated by the full ETL, changes by that one record's field and nothing else
(checked record by record against HEAD).

build.ts now fails if WASP-108 b is marked imaged. The count cannot see a false positive, and its
record carries neither a period nor an axis, so no separation check could. Control: the full ETL on
the old query (its cache still present) fails with "WASP-108 b is marked as imaged; it transits, and
only a companion star beside it was imaged (Bohn et al. 2020)."

The imaged sentence ended "and no map of it exists". Luhman 16 b, one of the 101 and a brown dwarf,
was mapped by Doppler imaging (Crossfield et al. 2014, Nature 505, 654). It now reads "and no map of
it is used here", which holds for all of them; the doc comment names Luhman 16 b. Control: the old
sentence fails "says a directly imaged exoplanet was seen as a point of light" only (1 failed, 843
passed of 844). Live on :4301: WASP-108 b and Kepler-22 b read "no image of this world exists";
Luhman 16 b and HR 8799 b read the new sentence.

48fd6fd missed a third copy of the claim it corrected: the body page's comment still said every
exoplanet gets a derived surface "since none has ever been imaged". It now says none has had its
surface imaged, as planet-appearance.ts and the README do.

A correction to 48fd6fd's message: HR 8799 b, c and d are Marois et al. 2008 (Science 322, 1348),
and e is Marois et al. 2010 (Nature 468, 1080), not "b to e (Marois et al. 2008)".

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

tools/etl/build.ts

@@ -438,10 +438,17 @@ function validateExoplanets(exoplanets: ExoplanetRecord[], starIds: Set<number>)
   console.log(`  ${withPeriod}/${exoplanets.length} have a measured period, ${withHostMass} a host star mass.`);
 
   // The planets photographed by direct imaging, whose card must not say no image of them exists.
-  // Measured: 102 of 102 flagged in the archive are in the catalogue. What this catches is the
-  // join by name failing, which would put every one of them back under "no image".
+  // Measured: 101 of 101 flagged in the archive, and not transiting, are in the catalogue. What this
+  // catches is the join by name failing, which would put every one of them back under "no image".
   const imaged = exoplanets.filter((exoplanet) => exoplanet.imaged).length;
   assertCondition(imaged >= MIN_IMAGED_EXOPLANETS, `Only ${imaged} exoplanets are marked as imaged (at least ${MIN_IMAGED_EXOPLANETS} expected).`);
+  // And the one the flag is wrong on, which the count cannot see: a 2.68-day transiting hot Jupiter
+  // 0.15 mas from its star, flagged for the companion star a survey imaged beside it. Its record
+  // carries neither a period nor an axis, so no separation check could catch it either.
+  assertCondition(
+    !exoplanets.some((exoplanet) => exoplanet.id === 'WASP-108 b' && exoplanet.imaged),
+    'WASP-108 b is marked as imaged; it transits, and only a companion star beside it was imaged (Bohn et al. 2020).'
+  );
   console.log(`  ${imaged} were imaged directly.`);
 }