[Nir-Az]

Overview: Test harness now waits 4 seconds after powering on a D585 Prototype device before tests start streaming, working around a FW window in which the accelerometer delivers no data.

Tracked on [RSDEV-13011]

Why

D585 Prototype FW produces no accelerometer data if streaming starts within the first ~5-7 seconds after power-on (RSDEV-13011). The pipeline aggregator publishes framesets only when every resolved stream has delivered, so any test whose config includes the IMU fails with Frame didn't arrive within 5000. In a full nightly with the D585 Prototype enabled this failed 4 tests (rec-play test_pipeline_interface, test_pipeline_start_stop, test_accel_streaming[100/200]) and retry-masked a fifth (depth-unit).

Summary

• module_device_setup (conftest): after the device-enable handshake, if an enabled device is a D585 Prototype ("D585 Proto" name match — D585S and production D585 are not affected and not delayed), sleep D585_BRINGUP_SETTLE_SEC = 4 (measured threshold ~3s + 1s margin) before yielding to tests. Both single- and multi-device paths.
• No cost for other SKUs; ~4s per D585 Prototype module (+23s on a full nightly).
• Temporary workaround — to be reverted once the FW fix for RSDEV-13011 is deployed (removal tracked in RSDSO-21766).

Test plan

• infra-tests: 148 passed, 0 failed (rerun after the name-gate narrowing)
• Full nightly (D585 Prototype included, D585S excluded), same bench as baseline: baseline = 4 failed + 5 retried; with this patch = 356 passed, 0 failed, 0 retried (all platforms green)
• Settle confirmed active in per-test logs ("D585 bring-up settle: waiting 4s before tests")

---

🤖 Generated by AI

[sysrsbuild-gh-agentic]

rs-agentic bot automated review — 3 inline comments below.