fix(android): cap interceptor body reads to prevent OOM on large responses

[imsankalp]

Summary

Fixes #17

• Android captureResponseBody: replaces source.request(Long.MAX_VALUE) + buffer.clone().readUtf8() with a bounded read (limit + 1 bytes to detect truncation without consuming it). Appends [... truncated] when the body exceeds the cap.
• Android captureRequestBody: writes body to a temporary Buffer, then reads only min(size, limit) bytes. Appends truncation marker when needed.
• Both methods skip binary content types (image/*, video/*, audio/*, application/octet-stream, application/zip, application/pdf, font/*) and return [binary content — not captured] instead of buffering bytes.
• NetworkToolsManager.maxBodyCaptureBytes defaults to 256 KB and is publicly settable from app code.
• iOS: promotes the static kNTMaxBodyBytes compile-time constant to a maxBodyCaptureBytes property on NetworkToolsManager, so it is now also configurable at runtime (iOS already had the cap logic; this makes it consistent with Android).
• Adds setMaxBodyCaptureBytes(bytes) to the TurboModule spec (Android + iOS), the legacy bridge module, and the TypeScript exports.
• NetworkMonitorProvider accepts a maxBodyCaptureBytes prop and forwards it to native via a useEffect.

Test plan

• Make a request to an endpoint returning a large JSON payload (> 256 KB) — confirm body is captured up to 256 KB with [... truncated] marker
• Make a request for an image URL — confirm responseBody is [binary content — not captured]
• Set maxBodyCaptureBytes={512 * 1024} on NetworkMonitorProvider — confirm larger payloads are now captured up to 512 KB
• Call setMaxBodyCaptureBytes(128 * 1024) directly — confirm cap takes effect on subsequent requests
• Verify no regression on normal text/JSON responses under the cap

🤖 Generated with Claude Code