bug#64404: 30.0.50; condition-case-unless-debug mishandles :success

[Mattias Engdegård]

> 2. M-: (condition-case-unless-debug nil 1 (:success 2)) RET
>   => 1
> 
> This can easily be fixed in condition-case-unless-debug, and worked
> around by using explicit 'debug' with condition-case, but I was
> wondering if we want to do anything differently in
> internal_lisp_condition_case or the byte-compiler (e.g. new warning)
> instead?

Good catch! I don't think we need to bother about the Lisp interpreter, but I 
added a byte-compiler warning for it in 59a350cb91.
Would you like to repair condition-case-unless-debug?