feat: send 0x71 image-data chunks with BLE write-without-response (+ deep sleep command)

[davelee98]

Summary

This PR contains two changes (two commits), with the BLE throughput change built on top of the deep-sleep command:

1. BLE Write Without Response for image-data chunks (3f1a0c1)

Switches DIRECT_WRITE_DATA (0x71) chunks from write-with-response to BLE Write Without Response, removing the ATT write-confirmation round-trip per chunk (~2× fewer round-trips; the largest win is over an ESPHome BLE proxy, where each ATT round-trip crosses the network).

• The per-chunk application ACK is still awaited, so exactly one write is in flight at a time and firmware flow control is fully preserved — no firmware change required. Both ESP32 (PROPERTY_WRITE_NR) and nRF (BLEWriteWithoutResponse) already advertise the property and ACK every 0x71 chunk.
• write_command() gains a response flag, threaded through _write() and opted into only by the two 0x71 loops (_send_data_chunks, _send_partial_chunks). All other commands keep write-with-response.
• The characteristic is probed for the write-without-response property and falls back to write-with-response when absent.

2. Deep sleep command 0x0052 (f78908c)

• CommandCode.DEEP_SLEEP = 0x0052 and build_deep_sleep_command()
• OpenDisplayDevice.deep_sleep(): sends 0x0052 and tolerates the link dropping during/after the command (ESP32 power-latch ACK-then-power-off, ESP32 no-latch immediate sleep, Silabs Flex ACK-then-close-and-EM4). A disconnect or missing ACK is treated as success; an explicit 0xFF52 not-supported error frame raises ProtocolError.
• PowerOption.deep_sleep_enabled: power_mode == BATTERY and deep_sleep_time_seconds > 0, mirroring the firmware sleep-entry condition.
• CLI: opendisplay sleep <device> subcommand.

Happy to split these into separate PRs if preferred.

Testing

• 533 unit tests pass (pytest tests/unit), including new tests for the write-without-response probe/fallback (tests/unit/test_connection_write_no_response.py) and the deep-sleep command across all firmware response shapes (tests/unit/test_device_deep_sleep.py).
• Verified on hardware via the Home Assistant integration: image sends over an ESPHome BLE proxy show the expected reduction in per-chunk latency, with firmware ACK flow control intact.

🤖 Generated with Claude Code

[davelee98]

write-without response was tested and working, but does not speed up transfers by itself. Deep sleep command is implemented but needs to be further enhanced with a "Sleep For X Seconds" to do what user intends. Otherwise if deep sleep is commanded on a target without a preconfigured sleep timer, it won't do anything on the firmware side. This update has some dependencies for the latest home assistant deep sleep integration updates.

[codecov]

Codecov Report

❌ Patch coverage is 60.00000% with 24 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/opendisplay/cli.py	0.00%	22 Missing ⚠️
src/opendisplay/device.py	92.00%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!