Focus Test: default the Method to "Z axis steps" when the machine has a Z axis #550

Split out from @atkaper's suggestion on #539 — the comment there asks for the "Z axis steps" method to become the default when the machine definition indicates it has a Z axis. Same idea, tracked separately so #539 stays about the Focus Test itself.

Would it be an idea to make the Method "Z- axis steps" the default if the machine definition indicates it supports a z-axis?

Most of this already exists

available_focus_modes() already does the capability filtering, and already orders the result by preference:

modes = [FocusMode.RAMP]
if machine is None or step_registry.get("CommandStep") is None:
    return modes
if not _driver_uses_gcode(machine):
    return modes
modes.insert(0, FocusMode.MANUAL)
if machine.has_z_axis:
    modes.insert(0, FocusMode.Z_AXIS)
return modes

So Z_AXIS is already first whenever it is available. has_z_axis is the right indicator too — it is Machine.axes.get(Axis.Z) is not None, i.e. a 2-axis laser simply omits Z.

The gap is only the dialog's initial selection

FocusTestDialog.__init__ seeds itself from the dataclass default instead of from that ordered list:

d = FocusTestParams()          # mode defaults to FocusMode.MANUAL
...
if d.mode in modes:
    self.mode_row.set_selected(modes.index(d.mode))

FocusTestParams.mode defaults to FocusMode.MANUAL, and MANUAL is in modes on every path that also reaches Z_AXIS, so the condition is always true and the dialog always opens on Manual — even on a Z-axis machine where Z_AXIS is offered and listed first. That is the whole of @atkaper's ask.

It also works by accident in the one case it handles correctly: with no machine, modes is [RAMP], MANUAL is not in it, nothing is selected, and the ComboRow falls back to index 0, which is RAMP.

Suggested shape

Default to the first available mode rather than to the dataclass value:

self.mode_row.set_selected(0)

since modes is already preference-ordered. That falls out correctly in all three cases:

Two notes

Leave the dataclass default alone. FocusTestParams() has no machine reference — it cannot — and it doubles as the source for every other field in that dialog (start_offset, step, count, …). Changing mode there would change the default for headless and scripted creation too, where the safe answer is still Manual. The capability decision belongs in the dialog, next to the modes list that already knows the answer.

Worth considering but out of scope here: the method could be greyed out with an explanatory row rather than merely absent when there is no Z axis, so a user on a 2-axis machine learns why it is not offered. available_focus_modes() already returns without it, and the command rejects a Z_AXIS params object with ValueError, so the current behaviour is safe — it is just not self-explanatory.

UI tests for the dialog live in builtin_addons/rayforge-addon-laser/tests/ui_gtk/test_calibration_test_dialogs.py; the backend behaviour of available_focus_modes() is already covered in tests/doceditor/test_calibration_tests.py.