Add CI, Makefile, README updates, Go 1.22 and AI-controlled paddle

[jenska]

Motivation

• Provide automated CI and development tooling for the project.
• Make it easier to run, build and test the game with standard make targets.
• Improve the single-player experience by replacing the naive computer paddle with a predictive AI.
• Update module settings and docs for modern Go and the correct engine name.

Description

• Add a GitHub Actions workflow /.github/workflows/ci.yml that checks formatting with gofmt and runs make test on ubuntu-latest, including installation of required native libs.
• Add a Makefile with common targets: run, build, test, tidy, and fmt.
• Update README.md to reflect Go 1.22+, use the Ebitengine name and provide run/install/make instructions and controls.
• Bump go version to 1.22 in go.mod and add several indirect module requirements.
• Replace the simple AI in game/paddle.go with a new predictive computer paddle implementation including updateComputer, nextAITargetY, predictBallY, isBallApproaching, randomError, and a clamp helper, plus configurable AI parameters and RNG seeding.

Testing

• The CI workflow will run gofmt -l main.go game/*.go and make test as automated checks.
• Local development targets were exercised with make fmt and go test ./..., and the tests completed successfully.
• Formatting verification is enforced in CI to prevent unformatted changes from merging.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 42259be0ec

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]

Mirror right-paddle X target in AI prediction

For the right-side computer paddle, targetX resolves to p.x here, but the actual contact/reset position for right-paddle hits in update is p.x - paddleWidth/2 - ballRadius. That 20px mismatch makes timeToReach too large, so predicted Y is systematically off; in AI-vs-AI mode this causes avoidable misses when the ball has meaningful vertical speed.

Useful? React with 👍 / 👎.