Replace demo benchmarks with captured synchronization and performance measurements

[sargunv]

Description

Replace the demo's projection-callback latency and Compose-frame drop estimates with repeatable camera animation, camera setter, and input-response scenarios. The same scene runs from the demo panel or capture runner, using local map content and a production placedAt overlay, with configurable Android presentation, FPS cap, and geometry load. Capture launches bypass normal demo state creation.

The runner retains raw artifacts and measures visible map/overlay separation. Android adds scheduled app CPU time, window GPU duration and missed deadlines, available driver GPU work periods, FrameTimeline events, and input-to-captured-display latency bounds. Desktop and iOS add process CPU counters, including runs without video. Chromium currently supports visual measurements only. Unavailable GPU/presentation data is not reported as zero; captured-display latency excludes the physical touchscreen and panel.

Performance-only results require a matching artifact/configuration/device visual capture whose raw pixels pass validation. Capture and lifecycle timeouts cover the full allowed startup, workload, and shutdown. Delete the previous benchmark workloads, tile-prefetch machinery, and timing estimators. Keep synthetic measurement tests in hygiene CI; the demo does not gain a unit-test suite.

Renderer mitigations stay outside this PR. This establishes their measurement harness for #407; selected designs are recorded in #1366, #1367, and #1368.

Validation

• Eighteen Python analyzer tests cover known pixel offsets, latency bounds, sparse input recordings, strict Android capture clocks, incomplete metric logs, rejection of mismatched or stationary visual references, frame-interval filtering, desktop artifact identity, low-FPS captures, external URLs, and Android animation-scale validation.
• Static checks and Android release build. A release capture verified 727 window frames within the trace interval from 795 collected reports, with none dropped. Android API 36 emulator capture with a deliberate 14-second launch delay retained the complete measurement interval; matched performance-only capture and reanalysis verified reference handling.
• Earlier harness validation included Android Lint, iOS 26.5 simulator, macOS Metal desktop, and Chromium/WebGL builds and captures, plus CPU-only runs on Android, desktop, and iOS. Forty matched mitigation/control captures used one verified artifact per platform with runtime experiment flags; those prototype changes are not included here, and comparisons need replication after the capture-launch isolation correction.
• No physical-phone GPU or touch-to-photon validation. GPU/deadline metrics and calibrated input capture remain Android-specific.

AI assistance

Implemented and investigated with OpenAI Codex (GPT-6).

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-09T03:21:28.351590Z	c54d098	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1fd056f4a0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d4804c66a0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 131b67315a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2eaf82badc

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".