Automate Android benchmark running / data processing

[louwers]

• Add one warmup run for caching data.
• Move slower so that we don't benchmark the renderer not rendering anything.
• Reduce the amount of styles / locations / total benchmark runs so that the benchmark runs quicker. This allows us to do multiple runs on different devices (of the same type) for more trustworthy data.
• Add scripts for kicking off benchmark, downloading / processing benchmark results.
• Add Python script for plotting benchmark results.

Usage:

Kick off benchmark (do this once per renderer, need to edit the script):

cd platform/android
scripts/run-benchmark.sh

From root directory. Collect benchmark results... (Can take multiple --runArn):

outputDir=...
node scripts/aws-device-farm/store-test-artifacts.mjs --customerArtifacts --outputDir  $outputDir --runArn $runArn
node scripts/aws-device-farm/collect-benchmark-outputs.mjs --inputDir $outputDir --outputDir ~/Downloads/run-processed

Now ~/Downloads/run-processed will contain a bunch of JSON files with the benchmark results.

Use Node.js 23 (nvm use 23), collect results in a SQLite database:

node --experimental-sqlite scripts/aws-device-farm/update-benchmark-db.mjs --dbPath benchmark.db --jsonDir ~/Downloads/run-processed

Install matplotlib, plot results:

python3 scripts/aws-device-farm/plot-android-benchmark-results.py benchmark.db

[github-actions]

Bloaty Results (iOS) 🐋

Compared to main

    FILE SIZE        VM SIZE    
 --------------  -------------- 
  [ = ]       0  [ = ]       0    TOTAL

Full report: https://maplibre-native.s3.eu-central-1.amazonaws.com/bloaty-results-ios/pr-2991-compared-to-main.txt