Macros: use zero_reference_position when using beacon true zero

[miklschmidt]

This prevents scanned meshes from being offset from the true zero position, which results in an unintended additional z_offset to be applied via the incorrect bed mesh.

[miklschmidt]

Great, there's no way to set zero_reference_position dynamically...

[miklschmidt]

This now does 2 things:

1. Adds a sanity check, to ensure zero_reference_position is set to safe_z_home coordinates.
2. Allows the user to enable model calibration each print. We might need to make this default, it seems there are issues when doing the contact wipe without at least zeroing via contact before probing, when switching between significantly different bed surfaces.

[miklschmidt]

Update to point 2: It now zeroes Z before doing the contact wipe, regardless if a new model would be calibrated or not. This is to avoid the issue with vastly different bed plate material thickness, the probed offset does not seem to work in that case - not sure why.

[HelgeKeck]

looks good to me