* Use less tap dance memory.
Use dynamically allocated sparse array for tap dance state, dynamically allocate tap dance state when needed and free it when the tap dance is done.
* new approach
* Use null, check for null
* Reformat with docker
* Use uint8 with idx rather than uint16 with keycode in state
* fix accidental change
* reformat
* Add null check
* add documentation tip suggested by tzarc
* Only allow tap dance state allocation on key down, not on key up
Co-authored-by: Sergey Vlasov <sigprof@gmail.com>
* Only allow tap dance allocation on key down, not on key up
Co-authored-by: Sergey Vlasov <sigprof@gmail.com>
* add user action required section
---------
Co-authored-by: Sergey Vlasov <sigprof@gmail.com>
