Add command to initialize a bilinear ABL mesh

[vehystrix]

Description

When flashing new firmware, the bed leveling mesh is lost in most cases. When you have the mesh saved somewhere (eg from the output of M503) in the form of a bunch of G29 W commands, it is not possible to directly restore the mesh. First you need to home XYZ and then you need to do a full G29 run.

With this PR, I added the new option G29 P<value> to allow you to initialize the mesh to the specified value, after which you can use G29 W commands to fill the mesh.

I also removed the requirement to home XYZ before using G29 W

Just like with G29 W, any values outside the -10 to 10 range are rejected

Old way to restore the mesh:

G28  ; takes a while depending on the print head position
G29  ; takes even longer
M104 S0  ; reset temperature after bed leveling
M140 S0  ; reset temperature after bed leveling
G29 W I0 J0 Z0.01
...
G29 W I6 J6 Z-0.02
M500

New way to restore the mesh

G29 P0
G29 W I0 J0 Z0.01
...
G29 W I6 J6 Z-0.02
M500

Requirements

AUTO_BED_LEVELING_BILINEAR

Example

Example

Send: M420 V
Recv: echo:Invalid mesh.
Recv: echo:Bed Leveling OFF
Recv: echo:Fade Height 10.00
Recv: ok

Send: G29 W I0 J0 Z0.01
Recv: Error:No bilinear grid
Recv: ok

Send: G29 P0
Recv: Bilinear Leveling Grid:
Recv:       0      1      2      3      4      5      6
Recv:  0 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  1 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  2 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  3 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  4 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  5 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  6 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv: 
Recv: ok

Send: G29 W I0 J0 Z0.01
Recv: X:-10.30 Y:0.00 Z:10.00 E:0.00 Count X:-819 Y:0 Z:4000
Recv: ok

Send: M420 V
Recv: Bilinear Leveling Grid:
Recv:       0      1      2      3      4      5      6
Recv:  0 +0.010 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  1 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  2 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  3 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  4 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  5 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv:  6 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000 +0.000
Recv: 
Recv: echo:Bed Leveling ON
Recv: echo:Fade Height 10.00
Recv: ok

Send: G29 P100
Recv: Error:Bad initialization value
Recv: ok

Send: G29 P0.5
Recv: Bilinear Leveling Grid:
Recv:       0      1      2      3      4      5      6
Recv:  0 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  1 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  2 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  3 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  4 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  5 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv:  6 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500 +0.500
Recv: 
Recv: ok

[vehystrix]

So I took a look at the changes you made here, and I don't think it'll work properly (I may have missed something and didn't actually built and tested it yet)

• G29 P doesn't seem to exit early so a full bed level is still done
• The mesh validity is determined by bedlevel.grid_spacing.x and this is no longer set

  Marlin/Marlin/src/feature/bedlevel/abl/bbl.h

  Line 58 in d0dc3b6

  static bool has_mesh() { return !!grid_spacing.x; }

• bedlevel.grid_start is similarly not set anymore

I guess replacing line 274 with the following would fix it, but this also kind of smells

      const float x_min = probe.min_x(), x_max = probe.max_x(),
                  y_min = probe.min_y(), y_max = probe.max_y();
      if (parser.seen('H')) {
        const int16_t size = (int16_t)parser.value_linear_units();
        abl.probe_position_lf.set(_MAX((X_CENTER) - size / 2, x_min), _MAX((Y_CENTER) - size / 2, y_min));
        abl.probe_position_rb.set(_MIN(abl.probe_position_lf.x + size, x_max), _MIN(abl.probe_position_lf.y + size, y_max));
      }
      else {
        abl.probe_position_lf.set(parser.linearval('L', x_min), parser.linearval('F', y_min));
        abl.probe_position_rb.set(parser.linearval('R', x_max), parser.linearval('B', y_max));
      }

      if (!probe.good_bounds(abl.probe_position_lf, abl.probe_position_rb)) {
        if (DEBUGGING(LEVELING)) {
          DEBUG_ECHOLNPGM("G29 L", abl.probe_position_lf.x, " R", abl.probe_position_rb.x,
                             " F", abl.probe_position_lf.y, " B", abl.probe_position_rb.y);
        }
        SERIAL_ECHOLNPGM(GCODE_ERR_MSG(" (L,R,F,B) out of bounds."));
        G29_RETURN(false, false);
      }

      // Probe at the points of a lattice grid
      abl.gridSpacing.set((abl.probe_position_rb.x - abl.probe_position_lf.x) / (abl.grid_points.x - 1),
                          (abl.probe_position_rb.y - abl.probe_position_lf.y) / (abl.grid_points.y - 1));
      
      bedlevel.fill(init_val);
      bedlevel.set_grid(abl.gridSpacing, abl.probe_position_lf);

      G29_RETURN(false, false);

[classicrocker883]

I use

  grid_start.x = MESH_MIN_X;
  grid_start.y = MESH_MIN_Y;
  grid_spacing.x = MESH_X_DIST;
  grid_spacing.y = MESH_Y_DIST; 

if that helps.