feature: add hebrew, arabic, hungarian text

[jottinger]

• This is a proof of a feature, not a change. It validates that GraphCompose can render Hebrew, Arabic, and Hungarian texts properly in a PDF by adding tests.

Part of #140

Summary

Type of change

• Bug fix (no public API change)
• New feature / public API addition
• Documentation only
• CI / build / tooling
• Refactor (no behavioural change)

Checklist

• PR targets develop (the integration branch); not main.
• Branch name follows <type>/<short-description> (e.g. feature/canvas-clip, fix/table-overflow, docs/recipe-themes); issue-prefixed names like 42/fix/short-description are also fine.
• ./mvnw -B -ntp clean verify passes locally.
• Java 17 compatible — no List.getFirst() / getLast(), no Thread.threadId(), no switch type / deconstruction patterns, no case null, default ->. CI matrices Temurin 17 / 21 / 25 and will catch JDK regressions.
• Public API change (if any): CHANGELOG.md entry added under the next ## v<X.Y.Z> — Planned heading.
• README touched (if any): DocumentationCoverageTest.readmeShouldUseCanonicalDslAndAvoidLegacyApis still passes — canonical fingerprints (GraphCompose.document(, DocumentSession, document.pageFlow(, BusinessTheme) preserved, no legacy tokens (GraphCompose.pdf(, PdfComposer, CvTemplateV1, …).
• Examples touched (if any): runnable via ./mvnw -f examples/pom.xml exec:java -Dexec.mainClass=…; if a new example, also added to GenerateAllExamples and the gallery row count in examples/README.md.

Linked issue

Part of #140

[DemchaAV]

Hello Joseph, thanks for this PR — the rendered output is a useful baseline validation.

I don't personally read Arabic or Hebrew, so I can't fully verify the semantic correctness of text order and shaping from the PDF alone.

From the current render:

• Hungarian accented text renders correctly
• Hebrew/Arabic glyphs are present

One caveat: the engine currently has no BiDi reordering or Arabic shaping (text is drawn in logical order), so anything that looks "connected/shaped" shouldn't be read as real shaping yet — I'd treat this as glyph coverage, not correct RTL.

So I'd like to land this as baseline Unicode / multilingual rendering coverage, not complete Arabic/Hebrew support.

Full Arabic/Hebrew support is tracked in #140 (I'm keeping it open), to be done with:

• RTL paragraph direction
• BiDi mixed text (Arabic/Hebrew + English + numbers)
• Arabic shaping / contextual forms
• line wrapping
• punctuation
• right alignment

Housekeeping I did on the PR so you're aware: retargeted the base from main to develop, changed Closes #140 → Part of #140 (so the merge keeps the tracking issue open), and pushed a small commit making the multiscript-font lookup cross-platform (added Windows font paths so the test runs on Windows/macOS/Linux instead of skipping).

Happy to merge this as an important first step. Thanks again!