fix(compare): correct four stale counts and four wrong prices

[datlechin]

Every comparison entry audited against the vendor's own site. The JavaScript-rendered pricing pages (JetBrains, DBeaver, Navicat) were driven through a real headless browser rather than guessed at.

Counts left over from the 18-database era

Postico promised "Postgres plus 17 other databases" and named five engines "and 13 more". Both read as 18 — on a page whose own comparison table said 25 two rows below. Same for HeidiSQL.

	before	after
postico.description	five named + 13	+ 20
postico.verdict	Postgres + 17	+ 24
postico.ogPunchline	Postgres + 17	+ 24
heidisql.longDescription	five named + 13	+ 20

Prices that contradicted their own entry

The OG card is generated from its own fields, so it drifts silently from the page it advertises — and nobody sees an OG image until it's already being shared.

	before	after
postico.ogCompetitorMetaHtml	$40	$69 — entry said $69/$99 throughout
datagrip.ogCompetitorMetaHtml	$100/yr	$99/yr — entry said $99 throughout

Prices now stale against the vendor

Both read off the rendered page:

	before	after	source
DataGrip	$9.90/mo	$10.90/mo	jetbrains.com/datagrip/buy renders "US $10.90 per month"
Navicat	$1,299 perpetual	$1,599 perpetual	navicat.com renders "1 Perpetual License(s) … USD 1,599.00"

Confirmed correct, left alone

• Postico $69 personal (3 devices) / $99 per device commercial — matches eggerapps.at
• Beekeeper $108/yr Indie — their page renders $9 / $14 / $18 per month

Deliberately NOT changed

DBeaver's Free / from $99/yr PRO is unverified. Their buy page renders prices only behind an async quantity selector; the figures that do render ($810/$410/$255/$155/$82) belong to Team Edition roles, not the desktop tiers. Rather than substitute a guess, it's left exactly as it was — flagging it here so it's on the record as the one open item.

DataGrip's $99 first year / $79 second / $59 from year three is likewise unverified — JetBrains' page only exposes monthly. Given monthly rose from $9.90 to $10.90, the annual figures may have moved too.

Two guards so this can't rot back

1. Relative counts must total 25. "plus N more" has to be re-derived every time the driver count moves, and nobody remembers to — which is how four of them ended up describing an app that no longer exists.

   Scoped to the sentence, because Postico's verdict names Postgres twice ("excellent for Postgres alone. TablePro covers the same ground for Postgres plus 24 other") and counting the whole string turns a correct 25 into a false 26. The first version of this test failed on exactly that and caught its own bug.

2. Any price on an OG card must appear elsewhere in the same entry — precisely the check $40 and $100 would have failed.

Verification

pint:  passed
Tests: 166 passed (2132 assertions)   ← REQUIRE_SSR=1, no skips

Four affected OG images regenerated and visually inspected — Postico's now reads $69 and "Postgres plus 24 more", with its 25 DATABASES stat finally agreeing with its own punchline.