fix(camera): ToupTek software trigger ignored after a switch on a stopped stream

[maragall]

Problem

On a ToupTek camera, switching from Hardware Trigger to Software Trigger while the stream is stopped leaves the camera unable to take software-triggered frames. The trigger mode becomes software, but the trigger source stays on the hardware input, so Trigger() is ignored.

Record + Z-Stack (#564) stops the stream between its phases, so a run started from Hardware Trigger live view hits this. The recording completes, then the z-stack's first trigger times out (Timed out waiting ... for a frame, aborting acquisition) and no z-stack is written. This happened 3/3 times on an ITR3CMOS26000KMA.

Cause

The SDK moves the trigger source (IOCONTROLTYPE_TRIGGERSOURCE: 1 = GPIO0, 5 = software) along with OPTION_TRIGGER only while the stream runs. _set_acquisition_mode_imp sets the source explicitly for hardware trigger, but leaves it to the SDK for software trigger. Measured at camera level, with no Record + Z-Stack involved:

Sequence	Source after the switch	Software trigger
Hardware → Software, stream running	5	frame in 0.08 s
Hardware-trigger live, then Software, stream running	5	frame in 0.08 s
Hardware → Software, stream stopped	1	no frame
Same, with a continuous recording in between	1	no frame
Previous state, source set to 5 by hand	5	frame in 0.08 s

Fix

When entering SOFTWARE_TRIGGER, set the trigger source to software explicitly, the way the hardware branch sets its own source. A camera that rejects the call only logs a warning, which is the same behaviour as today. Only ToupTek cameras are affected, and only when they enter software trigger. At startup and on a live-view switch the source is already 5, so the write changes nothing there.

test_camera_toupcam_trigger_source.py models the stopped-stream behaviour with a fake SDK. It fails without the change and passes with it.

Verified on hardware

On the ITR3CMOS26000KMA:

• Every sequence in the table gives a frame.
• Hardware-trigger live still delivers 25–27 frames in 3 s.
• Wellplate acquisitions in Software and Hardware Trigger save every image, with identical results with and without the change.
• feat: Record + Z-Stack acquisition mode #564's 11 hardware tests all pass, including Hardware-trigger live (2/2). Details: feat: Record + Z-Stack acquisition mode #564 (comment)

Not tested: level-mode hardware trigger (this microscope uses edge mode) and other ToupTek models.

Stacking

This PR is based on feat/record-zstack-acquisition (#564) and merges into it; #564 is the branch that merges to master. The fix was briefly pushed directly on #564 (eb441dd) and has been reverted there (ee706e1), so this PR now holds it alone.

🤖 Generated with Claude Code

[maragall]

Closing: we are fixing this inside #564 instead (the z-stack switches to software trigger after its stream starts), so the shared ToupTek driver stays unchanged.

?? Generated with Claude Code