build: update Paparazzi to 2.0.0-alpha05

[AlvaroBrey]

Part of the AGP 9 upgrade pre-work. Paparazzi alpha02 cannot run on Gradle 9 in two separate ways, so it has to move before the AGP/Gradle bump.

Agent description

Motivation

Part of the AGP 9 upgrade, split into small independently-shippable PRs. Paparazzi is a hard
prerequisite for Gradle 9, for the two reasons in the summary above.

The AGP-9 support claim needed checking, because the sources conflict. The alpha05 changelog
says "This release supports pre-AGP 9.0 consumers", which reads like AGP 9 is unsupported, and
the PR carrying the AGP-9 API migration (cashapp/paparazzi#2106, "[DNM] AGP 9.0 API Changes") was
closed unmerged. Resolved it against the 2.0.0-alpha05 tag's actual source instead:

• PaparazziPlugin.kt imports com.android.build.api.dsl.CommonExtension and contains no
  BaseExtension reference at all, so the migration did land, just via a different change.
• It contains isAgpAtLeast(major = 9), used to enable ASM instrumentation on AGP 9+: an
  AGP-9-aware feature branch, not a version cap.
• Issue [Paywalls V2] Add CarouselComponentView #2095 is closed / completed, milestoned 2.0.0-alpha05.

So alpha05 supports AGP 9, and the changelog line means it also retains pre-AGP-9 support.

On the golden images. alpha05 bumps LayoutLib 16.1.1 → 16.2.1 and Compose to 1.11.2, which can
legitimately shift rendering, so the two committed ui/debugview PNGs were the main risk. What the
measurements showed:

	notConfigured	configured
alpha02 (baseline, local macOS)	0.047221%	0.455492%
alpha05 (local macOS)	0.055340%	0.467417%

Both already fail locally on unmodified main by the first row's margins, while
ci/circleci: test_defaults_debug is green on main (that job runs testDefaultsDebugUnitTest
across all modules, which includes DebugViewSnapshotTest). So the goldens match CI's Linux
rendering and simply do not match macOS. alpha05 adds roughly +0.008% / +0.012% on top of that
pre-existing gap, which is the same order as the environment noise itself.

Re-recording on macOS would therefore replace CI-valid images with locally-valid ones and break the
job that currently passes. Leaving them untouched; CI is the authority here, and if alpha05's
rendering change does cross the threshold on Linux, test_defaults_debug on this PR will say so.

Description

• gradle/libs.versions.toml: paparrazzi (sic) 2.0.0-alpha02 → 2.0.0-alpha05, and removed the
  now-false # Can't use alpha04 because it requires higher Kotlin version than we're using
  comment, which this PR's Kotlin dependency resolves.

Also checked while here, no changes needed:

• Paparazzi task names are unchanged between alpha02 and alpha05, and
  fastlane/Fastfile's <module>:recordPaparazziDebug (which has no flavor, while both modules
  carry the apis dimension) resolves correctly to recordPaparazziDefaultsDebug via Gradle's
  camel-case abbreviation. Confirmed with --dry-run.
• fastlane/Fastfile carries a commented-out paparazzi.snapshot.dir property noting it is
  unsupported in the current version. Left alone; not required here and worth its own look.

Tested: build configures cleanly on Kotlin 2.2.10 with no stdlib conflict;
:ui:revenuecatui:testDefaultsDebugUnitTest 1678/1679 pass, the single failure being the
pre-existing PaywallComponentsTemplatePreviewRecorder error from an uninitialized
upstream/paywall-preview-resources submodule in this environment; detektAll clean;
:ui:debugview:recordPaparazziDefaultsDebug still succeeds, so the path CI actually uses works
(re-recorded output was reverted, goldens are byte-identical to main).

[emerge-tools]

📸 Snapshot Test

31 modified, 590 unchanged

Name	Added	Removed	Modified	Renamed	Unchanged	Errored	Approval
TestPurchasesUIAndroidCompatibility Paparazzi com.revenuecat.testpurchasesuiandroidcompatibility.paparazzi	0	0	31	0	251	0	✅ Approved
TestPurchasesUIAndroidCompatibility com.revenuecat.testpurchasesuiandroidcompatibility	0	0	0	0	339	0	N/A

---

🛸 Powered by Emerge Tools

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.72%. Comparing base (1382809) to head (1c2e5a2).

Additional details and impacted files

@@                   Coverage Diff                   @@
##           build/bump-kotlin-poko    #3937   +/-   ##
=======================================================
  Coverage                   82.72%   82.72%           
=======================================================
  Files                         425      425           
  Lines                       17472    17472           
  Branches                     2629     2629           
=======================================================
  Hits                        14453    14453           
  Misses                       2132     2132           
  Partials                      887      887           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[AlvaroBrey]

See #3934 (comment)