perf(batch-lint): batch file processing to reduce memory usage

[luojiyin1987]

Closes #48

Problem

Current batchLint reads all Markdown files into memory before processing, and each worker call handles only 1 file. For large repositories, this causes high memory usage and excessive IPC overhead.

Changes

• Process files in batches of concurrency size (matching thread count)
• Each batch reads files and sends them to a single worker call (1 IPC per batch instead of N)
• Memory bounded by concurrency files at a time instead of all files
• Add batch content list test for lint-worker

Verification

• npm test: 26/26 passed
• npm run build: successful