feat(spm): complete iOS SPM migration — remove all podspecs, use arrrrny/google-mlkit-swiftpm

[arrrrny]

Summary

Complete the iOS SPM migration by removing all CocoaPods podspecs and transitioning every plugin to pure Swift Package Manager, backed by arrrrny/google-mlkit-swiftpm (release 9.0.0-1).

Changes

Core migration

• Delete all 22 .podspec files — no more CocoaPods for any plugin
• Update 11 Package.swift files — all ML Kit plugins now reference arrrrny/google-mlkit-swiftpm instead of d-date/google-mlkit-swiftpm
• Update Package.resolved files in example apps

Text recognition fixes (d-date/google-mlkit-swiftpm#106)

• Replace monolithic MLKitTextRecognitionResources.bundle with 5 per-language OCR bundles: Latin, Chinese, Japanese, Korean, Devanagari
• These bundles are included in the consuming app's Xcode project via Copy Bundle Resources build phase
• -ObjC linker flag added to prevent Objective-C category stripping (fixes unrecognized selector crashes in MLKitCommon)

Build infrastructure

• .gitignore updated for SPM build artifacts (**/.build/, **/.swiftpm/, per-package Package.resolved)
• Stale artifacts cleaned: MLKitTextRecognitionCommon.full.dylib.gz, .build directories, old Package.resolved files

New example app

• packages/text_recognition_example/ — minimal focused example that generates a test image with Latin text and runs ML Kit OCR, verifying the full pipeline end-to-end

Root causes fixed

1. Invalid model path — MLKitTextRecognitionCommon shipped as Mach-O object file; SPM stripped model data sections. Fixed by shipping as static .a archive (via alex-pan-invos's textfix5 → now self-hosted on arrrrny)
2. unrecognized selector crash — Objective-C categories in MLKitCommon stripped by linker. Fixed by adding -ObjC / -all_load linker flags
3. Resource bundles not found — LatinOCRResources.bundle must be in main app bundle (since bundleForClass: returns main bundle with static linking). Fixed by adding bundles to app target's Copy Bundle Resources

Testing

• Verified on device (iPhone, iOS 18.7.1): text recognition initializes and processes images successfully
• Main example app builds and runs with camera + text recognition + all ML Kit modules

[arrrrny]

Thanks to @alex-pan-invos
d-date/google-mlkit-swiftpm#106 (comment)

[arrrrny]

I tested text recognition on my phone and it works, and barcode scanner already works. I did not test others but since these are extremely important, this can be a good jump start and for people who need barcode and or text recognition only, this is fully spm tested.

[pumano]

Would love to get this landed. Because after upgrading from old mac to fresh I can't use that lib anymore.

Latest error when use develop branch:

Failed to build iOS app
Swift Compiler Error (Xcode): Type 'VisionImage' has no member 'visionImage'
/Users/userhere/.pub-cache/git/google_ml_kit_flutter-8cb0e47bec1fa58704c4043284738dd25576884a/packages/google_mlkit_text_recognition/ios/Classes/GoogleMlKitTextRecognitionPlugin.swift:86:34

[fbernaly]

@arrrrny : make the PR against this branch: https://github.com/flutter-ml/google_ml_kit_flutter/commits/feature/spm/

[fbernaly]

is this replaced by #881?

[arrrrny]

Yes, this PR is replaced by #881 which targets feature/spm as you requested. Closing this in favor of #881.