Fix for #1518 custom functions bound to "0" break motion parsing

[erganemic]

This PR should fix #1518.

This PR gets rid of evil-redirect-digit-argument, and replaces its effects with a custom variable. The variable is a list of functions that can be bound to digits (always 0 in practice, I'd imagine, but I'm sure someone's workflow would be affected). This allows for greater consistency in how functions can be bound to 0 - just use any of the define-key variants like normal, then add the relevant function to the custom variable, instead of binding with evil-redirect-digit-argument.

It passes all tests, except for the ones that explicitly rely on "0" being bound to the wrapper function for evil-beginning-of-line, which I've fixed.

One thing I'd like feedback on is that I don't like the way that the new evil-digit-bound-motions custom variable is set with customize-set-variable when evil-respect-visual-line-mode is enabled. Since I typically setq custom variables in my init.el instead of using a custom file, I don't know if setting this one programmatically will cause issues in general.

This PR also removes evil-beginning-of-line-or-digit-argument, since it hasn't been touched since 2011.

[tomdl89]

Awesome! Didn't expect an issue and a PR! I'm on hols right now, and no one else is much active around here, so don't be disheartened if you don't get feedback or merged soon. Shout/pester if you get impatient. I'll try to take a look in a week or two.

[tomdl89]

Just took a look and it looks (suspiciously) simple, so great job if my suspicions are unfounded. One thing I noticed is that ln78 of evil-keybindings still has a reference to evil-digit-argument-or-evil-beginning-of-line so pressing 0 when in Info mode now doesn't work. Otherwise, I'll need to spend a bit longer trying to break it before giving a 👍.

[erganemic]

@tomdl89 Awesome, glad to see this pulled in! I guess my concern about a custom variable to maintain a list of digit-bound motions was unfounded? I'm super new to Emacs package development, so I wasn't sure if this solution was the best way to provide other packages in the ecosystem with an interface for binding their own motions to digits or not.

My concern was that while the old behavior was actively harmful (breaking motions with a 0 in them if you tried rebinding another command to 0), the new behavior might not be that useful, since external packages wouldn't have an easy way of setting the custom variable.

If you think that they'd have an avenue for doing that, great! If not, I can keep looking into different ways to accomplish this behavior - maybe using cl-digit-char-p in the keypress parser in order to determine if a given command is bound to a digit?