NOZZLE_CLEAN_FEATURE is no longer dependent of HAS_BED_PROBE

[jbrazio]

Followup #4054 .
This PR changes NOZZLE_CLEAN_FEATUREso it is no longer dependent of HAS_BED_PROBE.

[thinkyhead]

Does declaring a function as inline in a header without including the implementation still allow the compiler to make it inline? Obviously not outside of Marlin_main.cpp… but within that file will it still inline the function sometimes, even though it is now exposed to the linker?

[thinkyhead]

I'm going with this in Marlin.h instead, to save on program space:

/**
 * Blocking movement and shorthand functions
 */
static void do_blocking_move_to(float x, float y, float z, float fr_mm_m=0.0);
static void do_blocking_move_to_axis_pos(AxisEnum axis, float where, float fr_mm_m=0.0);
static void do_blocking_move_to_x(float x, float fr_mm_m=0.0);
static void do_blocking_move_to_y(float y);
static void do_blocking_move_to_z(float z, float fr_mm_m=0.0);
static void do_blocking_move_to_xy(float x, float y, float fr_mm_m=0.0);

[Blue-Marlin]

For me this seemed always to be a bit random. My best description would be:
If the 'compiler' is able to inline the function everywhere it does not produce a function we can call from outside. This is later realised by the 'linker' and we get an error.

[thinkyhead]

The "shorthand" functions, we presume, will be inlined because all they do is pass-through their arguments. A compiler should catch that right away and not generate two function calls.

[jbrazio]

Does declaring a function as inline in a header without including the implementation still allow the compiler to make it inline

I would think so, otherwise gcc would bark at us for sure for violating some C spec.
-edit- in fact this is one way to do it on classes.

And we must never forget that with -O2 the compiler is far more smarter than us concerning inlining, this means that even if we want to force something to inline is ultimately a compiler decision what to do.

[jbrazio]

I kept out void do_blocking_move_to(float x, float y, float z, float fr_mm_m=0.0); because it was static and I didn't want to get someone angry at me. :-D

[thinkyhead]

This isn't really performance-intensive code, so reducing size is more important. Inline functions of course cost more storage space. Compiling with NOZZLE_CLEAN_FEATURE enabled…

• With simple inline specifiers on all: 99,728
• Adding FORCE_INLINE to all: 100,020 (+292)
• Define do_blocking_move_to_axis_pos static: 99,764 (+36)
• Define all as static in the header file: 99,692 (-36)
• Declare in header, but define in cpp file: 99,732 (+4)
• Declare static in header, define in .cpp: 99,474 (-254)

Apparently declaring them static in the header while defining them in the .cpp file is the cheapest option, size-wise. The downside is one or two more call-hops when we use these shorthand functions.