perf: optimize Protocol V2 BLE transfer throughput and metrics

[wabicai]

Summary

• Prefer Protocol V2 BLE writes without response across React Native, Electron, and the CLI while retaining explicit with-response fallback and caller overrides.
• Size Protocol V2 BLE packets from negotiated platform capacity, keep the generic filesystem limit conservative, and use the verified 1960-byte budget only for fixed firmware and wallpaper package paths.
• Improve lossless Pro2 host-asset LZ4 block compression with a firmware-safe fallback for incompressible input.
• Report firmware transfer progress as whole-batch confirmed bytes, average bytes per second, and elapsed time across resources, components, verification, recovery, and retry delays.
• Preserve V2-first protocol acquisition for V2-only methods and document the tested BLE throughput and compatibility boundaries.

Measured results

• Wallpaper baseline (64 B transport packets / 1800 B file chunks): 8.81 KiB/s.
• Optimized wallpaper (244 B negotiated packets / 1960 B fixed-path chunks): 16.72 KiB/s average.
• Pro2 4B8F bootloader update (2,440,562 B): 145.62 s transfer, 16.37 KiB/s average, 13.65 s install, 211.96 s CLI end-to-end, 0 retries.
• The 15 ms firmware connection interval measured 20.3 KiB/s versus 14.22 KiB/s at 30 ms. Connection-parameter changes remain outside this SDK-only PR.

Test coverage

• 10 focused Jest suites passed: 528 passed, 4 skipped, 532 total.
• yarn workspace @onekeyfe/hd-core build passed.
• Targeted ESLint for FirmwareUpdateV4.ts passed.
• Coverage includes packet capacity/fallback behavior, write mode selection, V2 frame boundaries, file-write retry/progress, V2-only connection acquisition, host-package integrity and incompressible fallback, and whole-batch firmware metrics.

Pre-landing review

• No P0/P1 findings after tracing the file-transfer, retry, MTU, compression, and progress-event paths.
• git diff --check passed.
• The full @onekeyfe/hd-core package lint remains unusable because the existing package TypeScript configuration excludes test/dist inputs and reports unrelated baseline issues; targeted lint and the build are green.

Documentation

• Updated docs/testing/pro2-ble-performance.md with the CLI commands, device/firmware versions, before/after results, bottleneck analysis, safe limits, platform behavior, and the firmware metric contract.

Companion PR

• App integration Draft PR: feat: surface firmware update speed, ETA, duration, and failure types app-monorepo#13025
• The App PR consumes these firmware transfer metrics for cross-platform speed/ETA UI and firmware-update analytics. Its current SDK dependency remains backward-compatible until this SDK change is published and bumped separately.

Test plan

• Verify wallpaper upload over BLE with negotiated packet capacity and a 1960-byte fixed-path chunk.
• Verify a locally built firmware update in bootloader mode on Pro2 4B8F.
• Verify transfer metrics remain cumulative across a Protocol V2 firmware batch.
• Verify React Native, Electron, CLI, low-level, transport, and core regression suites.