"Magic numbers" are frowned upon, and if the length of the array changes this has to be changed in multiple hard-to-locate places. Please add a #define to replace the literal number 13 so that it can be changed in only one place.

@thinkyhead You have 1-point, 4-point and 7-point calibration; I've added the average of 2 x 7-point with common centre point -> 13-point for more precision. So 1, 4, 7, 13 are the magic numbers for calibration and should not be changed.

So I'm not in favor just to add a #define NUMBER_OF_CALIBRATION_POINTS 13 unless a comment is added : "DO NOT CHANGE"; it would give the false impression that changing the number of calibration points is as easy as changing the #define statement. It requires also changes to the calibration- and autotune-matrices, the steps in the for loops, etc...

However maybe we can consider to make a LOOP_CAL_PT(axis, start, step) similar to LOOP_XYZ(axis) to replace the for loops with the magic number 12, and a CalEnum for the specific magic numbers used all over the code.

See #8171 and #8173 for full implementation; I also changed the probe grids a bit to make them more universal. No change to the documentation is needed.
All magic numbers (array size, indices matrices, steps,...) are set by _7P_STEP. I tested it with 7p- (current set up - _7P_STEP = 1) and 13p-arrays (old set up _7P_STEP = 2). Of course when you extend the array one should do something (average or extend the calibration matrices used) to include the new probe points in the calculations (I left the extra code to average the 13p-array to a 7p-array in comments).