[dajiaohuang]

Fixes IMAGING-387.

What

DataReaderStrips.readImageData computes the uncompressed strip byte count as a long, but previously narrowed it to int before decompression. For the issue's 1-bit TIFF dimensions, 28,994,135,625 bytes becomes a negative integer; allocation then fails with NegativeArraySizeException instead of the documented imaging exception.

Validate the long request with Allocator.check before narrowing, then pass that checked size to decompression. Apply the same checked size to planar strip allocation. This rejects requests beyond the allocator's supported limit before any int truncation.

Tests

• DataReaderStripsTest checks that 28,994,135,625 is rejected and that a valid strip size is returned unchanged.
• Targeted DataReaderStripsTest: 3 tests, 0 failures/errors.
• Default Maven build (mvn): BUILD SUCCESS; 1,261 tests, 0 failures, 0 errors, 7 skipped. The task-local checkout was normalized to the repository's LF source convention for Checkstyle.

Contribution checklist

• Read the contribution guidelines.
• Read the ASF Generative Tooling Guidance.
• AI use: OpenAI Codex (gpt-6-luna) assisted with source analysis, implementation, and regression tests. Human operator: dajiaohuang.
• Ran the default Maven build successfully.
• Added behavioral regression tests.
• This description explains the issue and change.
• Commit subject and body identify the issue and summarize the change.