Replace obsolete forms

[Boruch-Baum]

• There's nothing "bad" about make-variable-buffer-local, but now that
  emacs has had defvar-local for such a long time, there's benefit in
  preferring it for readability and maintainability

[emacs-w3m-notifier]

I also see usage of function 'inline' whichsee docstring explicitly says "You don't need this." w3m.el 4722: (inline (w3m-cache-remove-1 url))))) w3m-util.el 1243: (when (timerp timer) (inline (cancel-timer timer)) t)) Should I try to 'fix' those, also? BTW: This was my first success using magit to create a PR into emacs-w3m. (it took a while and had some scary episodes because my magit configuration was probably non-standard). However, I don't see how to use magit to post comments, so I'm relying on the github mail feature.

…

-- hkp://keys.gnupg.net CA45 09B5 5351 7C11 A9D1 7286 0036 9E45 1595 8BC0

[yamaoka]

I also see usage of function 'inline' whichsee docstring explicitly says "You don't need this."

Yes, you don't need this if running the ELisp code without byte-compiling.　But when the code is compiled into elc, a function surrounded in inline will be replaced in the byte-code with its function definition itself, while compiled into the code equivalent to (funcall function-name) if it is not inlined.

(byte-compile
 (lambda (url)
   (inline (w3m-cache-remove-1 url))))
 => #[257 "\211\302^A^H\"\303\211^B\205/^@\304^C@   \"\210^BA@\211\262^C^CAA\211\26\
2^C|\210^A\303\211\223\210\211\303\211\223\210\305^C^H\"\211^P\207" [w3m-cache-\
articles w3m-cache-hashtb assoc nil remhash delq] 8 "                           
                                                                                
(fn URL)"]

(byte-compile
 (lambda (url)
   (w3m-cache-remove-1 url)))
 => #[257 "\300^A!\207" [w3m-cache-remove-1] 3 "                                    
                                                                                
(fn URL)"]

Using inline makes a byte-code faster, so we use inline if the code often or repeatedly runs.