bed_mesh: manual probing QoL improvements

[Artefact2]

These short commits improve things a little bit when manually probing for a bed mesh.

• Allow restarting a calibration sequence midway, without klipper panicking to death trying to redefine the NEXT command.
• Be more consistent in toolhead movements while doing the probing; moving to the first probe point did not respect horizontal_move_z by moving vertically like the other points.
• Allow saving and restoring the mesh with the MESH_BED_LOAD command. This makes it possible to reuse a previous calibration on multiple prints without having to re-probe (very useful if you probe manually). The command can be put, eg in the slicer's starting g-code section.

Signed-off-by: Romain “Artefact2” Dal Maso artefact2@gmail.com

[Arksine]

Just a couple of questions/comments from my perspective:

• Is there an issue I'm missing with moving the tool directly to the first coordinate at (X,Y,horizontal_move_z) vs lifting z first? Or just for the sake of consistency?

• I'm currently working on saving/loading state using Kevin's work-save_config branch. I should have something working soon. My plan is to use the current BED_MESH_CALIBRATE gcode and add two parameters.
  BED_MESH_CALIBRATE STORE=<profile_name>
BED_MESH_CALIBRATE LOAD=<profile_name>
  If no parameters are provided BED_MESH_CALIBRATE will probe and generate the mesh without accessing persistent storage. If STORE is provided it will probe the bed, then use save_config to write the data to a new section [bed_mesh profile_name], or overwrite that section if it exists. If LOAD is provided it will load the profile from persistent memory if it exists, or error out of it doesnt. This would allow users to store profiles at different temperatures or with different parameters.

[Artefact2]

Is there an issue I'm missing with moving the tool directly to the first coordinate at (X,Y,horizontal_move_z) vs lifting z first? Or just for the sake of consistency?

Yes, it solved a real issue for me. My printer homes outside the print area, and the bed is held in place with simple paper binder clips. Traveling from the homed position without lifting Z first crashes into the clips. (Because starting a calibration re-homes, you can't lift Z beforehand.)

[KevinOConnor]

Thanks.

I added your signed-off-by line to the commit message of the first patch (unregister NEXT) and committed it to the master branch.

The second patch (z lift before first probe) looks conceptually good, but (unless I'm missing something) it leaves the head close to the bed after the last probe. I think the z lift should still be present at the end. Perhaps we could abstract out a z_lift() method in the probe class and call it where needed.

The third patch (save/restore mesh) I think should wait for @Arksine . I'm definitely interested in seeing what we can come up with for automated save/restore mechanism.

-Kevin