Add Eject Option To Filament Load Message

[sarusani]

Add a third option ("eject") to the "Filament extruding & with correct color?" message.

Closes #4643

The third_choice text was rendered incorrectly. (Only showed the first letter of the text because the position was hardcoded.)
I made sure the third_choice text will be offset to the left, but not overlap with the second_choice text.

This uses quite a lot of mem, any help in optimizing is appreciated 😸

[github-actions]

All values in bytes. Δ Delta to base

Target	ΔFlash	ΔSRAM	Used Flash	Used SRAM	Free Flash	Free SRAM
MK3S_MULTILANG	228	0	247412	5653	6540	2539
MK3_MULTILANG	228	0	246690	5662	7262	2530

[gudnimg]

I wonder if we can save memory by converting the following arguments into a single struct and adding third_col parameter. And that way force the caller to specify the column. 🤔

void lcd_show_choices_prompt_P(uint8_t selected, const char *first_choice, const char *second_choice, uint8_t second_col, const char *third_choice)

[3d-gussner]

I wonder if we can save memory by converting the following arguments into a single struct and adding third_col parameter. And that way force the caller to specify the column. 🤔

void lcd_show_choices_prompt_P(uint8_t selected, const char *first_choice, const char *second_choice, uint8_t second_col, const char *third_choice)

The biggest cost +196 bytes is https://github.com/prusa3d/Prusa-Firmware/pull/4644/files#diff-5fbfdca43c192573e31d8e78d255c510fb6e213b118cf38173277ec2a2163880R2261-R2269

The last part only costs additional +32 bytes https://github.com/prusa3d/Prusa-Firmware/pull/4644/files#diff-5fbfdca43c192573e31d8e78d255c510fb6e213b118cf38173277ec2a2163880R3047-R3053

[gudnimg]

The biggest cost +196 bytes is https://github.com/prusa3d/Prusa-Firmware/pull/4644/files#diff-5fbfdca43c192573e31d8e78d255c510fb6e213b118cf38173277ec2a2163880R2261-R2269

In that case I wonder if some of the function calls are being inlined. You can try adding __attribute__((noinline)) to the definition and see if that helps. None of these are time critical.

[gudnimg]

I looked a bit into this flash usage, and I suspect this is due to the compiler not able to optimise the code like it could before. third_choice was always set to nullptr before, but now you give it a value which complicates things. I've pretty much ruled out inlining.

[3d-gussner]

@gudnimg Any idea if that can be optimized?

[3d-gussner]

@sarusani Please rebase the PR and solve the conflicts.

[3d-gussner]

Tested with MK404 ✔️

[3d-gussner]

@gudnimg Do you think we can save any bytes here? If not please approve.