Optimisation pass

[ASleepyCat]

Optimisation pass, primarily aimed at reducing the number of RGB update calls and reducing code size.
Mostly targets IIDX mode due to TT misfiring/misinputs from low polling rates, SDVX mode doesn't really have much going on except for the ADC stuff.

Benchmarks

Binary size

Using -O2. Compiled with avr-gcc 14.1.0 using pre-compiled builds from here. Sizes obtained by running avr-objdump -Pmem-usage beef.elf.

Section	master	optimisation-pass	Change
Program	21690	20954	-3.4%
Data	1146	1003	-13.5%

Polling rate

Tested by using Gamepadla and having an incrementing counter on the X joystick axis, averaged over five tests. Using 6000 samples. LED refresh rate is set to 60Hz.

Lighting mode	master	optimisation-pass	Change
HID (standby)	950.692	961.522	1.1%
HID (in-game)	959.552	998.004	4%
Off*	958.764	997.596	4%

*Off refers to the off lighting mode, not the "disable LED" toggle.

The fact that we see a 1% uplift in HID (standby) mode likely shows a real optimisation in our code, rather than it just coming from reducing the number of times we update the LEDs.

Changes

• Only update RGBs when absolutely necessary
  • This fixes cases where the Disable RGB modes performs worse than disabling all LEDs
  • Should be an overall performance improvement, especially on more static RGB lighting animations
• Change various ints to smaller sizes, especially around timing-related code
• Remove parameters that pass in global variables
• Remove some branching code
• Reduce default BEEF_LED_REFRESH value from 240 to 60
  • A compiler warning is now produced when setting this to a higher value
• Disable respective RGB update functions if RING_LIGHT_LEDS/LIGHT_BAR_LEDS are set to 0
  • Also whoops the LED counts weren't actually being used during compilation
• Optimise and refactor Ticker, BreathingPattern classes to reduce code size and branching code

[HWXLR8]

Is this ready for review?

[ASleepyCat]

Is this ready for review?

No, I still need to do some formal benchmarking and record the results here. Also need to do some playtesting.

[ASleepyCat]

Updated the description with my benchmark results, overall very good performance uplifts. Getting basically true 1000Hz polling without having to disable all LEDs.

[HWXLR8]

Fantastic work. Tested and working on my phoenixwan (full speed w/ LEDs enabled), still awaiting test results from controllers with poorer sensor alignment, but will merge anyway since there is an objective increase in several metrics. Thanks for the PR!