more complete support for $ in C/C++ identifiers

[pskocik]

ctags seemed to have some support for $ in identifiers, like it would index $BAR from #define $BAR but not $name from extern int $name;. Similarly it wouldn't let me define replacement macros e.g., '-D$my$struct=struct and then have foo from $my$struct foo indexed. This minipatch appears to fix these issues.

[coveralls]

Coverage remained the same at 84.906% when pulling 8228c5f on pskocik:master into b60a7b9 on universal-ctags:master.

[pragmaware]

I'm not 100% sure about $ being standard-compliant as part of an identifier but gcc seems to support in certain configurations so I agree ctags should too. If you can fix the small parenthesis detail I'll merge this without adding my own overrides.

[pskocik]

Oh, it's definitely nonstandard. But it looked like universal ctags was already trying to support it:

parsers/cpreprocessor.h:

#define cppIsident(c)  (cppIsalnum(c)					\
						|| (c) == '_' || (c) == '$')

/*  Is the character valid as the first character of a C identifier?
 *  C++ allows '~' in destructors.
 *  VMS allows '$' in identifiers.
 */
#define cppIsalpha(c)   (cppIsascii(c) && isalpha(c))
#define cppIsident1(c)  (cppIsalpha(c)					\
						  || (c) == '_' || (c) == '~' || (c) == '$')

and I found myself wanting it supported fully, hence the patch.

(so I guess the if and while better be replaced with:
if ( !cppIsident1(*c) )
and
while (cppIsident(*c) )
(I pushed this to https://github.com/pskocik/ctags/tree/alternative_dollar))

Apologies for messing up the parenthesis there. I forced-pushed a fixed version.

[pragmaware]

Great! Thank you!