server: fix Windows CI flake from the log color reset

[ServeurpersoCom]

Overview

On Windows the child logs in color, and the logger emits its reset after the newline. That stray escape lands in front of cmd_child_to_router:state: on the shared pipe, the prefix check fails, and download_finished is logged instead of handled, leaving the model stuck downloading. Hence the intermittent test_router_delete_model failures.

Additional information

A leading newline on the command fixes it. Reproduced on Linux with LLAMA_ARG_LOG_COLORS=on, keeping the cached blob while the registry no longer knows the model. pytest unit/test_router.py passes 17/17.

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: YES

[ggerganov]

This is supposed to fix failures such as: https://github.com/ggml-org/llama.cpp/actions/runs/34582172702/job/103208051836#step:6:2054

[ServeurpersoCom]

Yes for the test_router_download_model failure in that job, its only command line is download_finished and it arrives with the reset glued in front of it. The test_router_delete_model failure in the same job is a different one though: there download_finished is handled correctly and the SSE event goes out, but DELETE /models still hangs the full 600s because the download child never closes its stdout, so the monitoring thread never joins. I open a separate PR for that one.

[ServeurpersoCom]

I sent my agent through every red Server run of the past week to sort the Windows flakes by root cause, here is what it found :

cause                              jobs (7d)  share of router  status

download_finished command lost        42           80%         PR #28747 <- this PR
behind the ANSI reset

DELETE 600s, exit command              5           10%         PR #28755
never sent to the child

DELETE 600s, exit acknowledged         5           10%         PR #28762
but EOF never reached

test_completion_unified               20            -          PR #28759
returning 500

build / infra, misc                   22            -          out of scope

#28755
#28759
#28762