[BioCam]

The Problem

Nothing stops a read, a z-scan or a grid scan from being sent while the CLARIOstar's drawer is out. Each moves the plate carrier under the optics, and what the firmware does with the carrier out is not known: no recorded run was sent that way, and the reader has no documented refusal for it. Measuring with the plate outside the reader, or the carrier driving in on its own, would each be wrong to rely on; the driver already reads the drawer bit in every status it polls before a command, and ignores it.

PR Content

• Measurement._wait_until_idle_with_drawer_in(): the idle wait send_command does before a command, done before a RUN (_unchecked_fw_run), a z-scan (FOCUS_WELL) and a grid scan (PLATE_MAP_SCAN), raising RuntimeError("the drawer is open; close it first") if the idle status shows the drawer out; nothing else is sent. The command then goes with when_idle=False.
• The frames sent are unchanged: every existing test passes with its recorded reply sequence, and send_command is untouched.
• Tests on the recorded drawer-open status: an absorbance read, a fluorescence read, a z-scan and a grid scan are refused after one status query; with the reader busy first, the drawer bit is read from the idle status the wait ends on.
• Filter detection (FILTER_SCAN) moves only the filter slides and is not guarded.

🤖 Generated with Claude Code