test: add 422+ unit tests across 19 modules

[jacob-cotten]

Summary

Thank you for building pdfplumber-rs — it's an impressive pure-Rust port and we've been using it extensively. We wanted to give back by contributing comprehensive test coverage.

This PR adds 422+ unit tests across 19 modules in pdfplumber-core and pdfplumber-parse, covering edge cases, boundary conditions, and correctness invariants:

• encoding.rs — glyph name resolution, StandardEncoding boundaries, FontEncoding, EncodingResolver 3-tier logic
• edges.rs — edge generation, orientation, degenerate paths
• search.rs — regex patterns, unicode, bbox union, anchored patterns
• dedupe.rs — tolerance boundaries, font/size blocking, output ordering
• bidi.rs — RTL/LTR detection, neutral chars, field preservation
• shapes.rs — orientation classification, flip_y, rect construction, CTM transforms
• annotation.rs — subtype roundtrip, equality semantics, bbox preservation
• struct_tree.rs — deep nesting, MCID extraction, child ordering
• page_regions.rs — unicode masking, thresholds, custom margins
• path.rs — builder lifecycle, rectangle segments, clone independence
• html.rs — heading levels, HTML escaping, bold/italic detection, list rendering
• color_space.rs — all color spaces, ICC delegation, indexed bounds
• standard_fonts.rs — all 14 standard fonts, monospace invariant, bbox sanity
• words.rs — word split boundaries, vertical text ordering, tolerance semantics, CJK
• text.rs, layout.rs, table.rs, cmap.rs — additional coverage

Bug fix included

Also fixes the word-split tolerance check from > to >= to match Python pdfplumber's semantics, and corrects 3 pre-existing test expectations for vertical text ordering.

Test plan

• All 1,804 tests pass (cargo test -p pdfplumber-core -p pdfplumber-parse --lib)
• Zero ignored tests
• No functional changes beyond the tolerance boundary fix

Thank you again for your work on this project. 🙏

🤖 Generated with Claude Code