Remove redefinition of static keyword

[dvogel]

Problem: Ancient preprocessor hack could lead to very confusing build errors.
Solution: Use simpler XPM that reflects current source control approach.

My primary concern here is removing the redefinition of static. This was seemingly originally used as a workaround to GTKv1 vs GTKv2 API difference. When GTKv1 support was dropped, this redefinition should have also been abandoned in favor of declaring the XPM data static const. Those XPM files have been in version control, unchanged since 2004. Beyond being unnecessary, if one misplaces an #if or an #end this can result in very confusing "invalid storage class for function" errors because some functions will effectively be declared static const.

I've also removed this magick define/undef sequence because AFAICT it serves no purpose. If anyone knows a good reason this should remain, I'm open to keeping it but if we keep it then it's purpose should be documented. My best guess is that either at some point prior to the XPM files being put under version control they included code that interacted with that magick symbol. I consulted the v6.4 source code on the FTP archive and that version does not include any of this code so there seems to be a gap between v6.4 and the v7 sources in the git repo.

[chrisbra]

Thanks. I have no idea if this breaks for anybody. Best to include and see if anybody complains, but I doubt anybody will notice.

[chrisbra]

This change seems to cause some warnings: https://groups.google.com/g/vim_dev/c/1QH6AEZxqRw/m/awOu4EjeDgAJ

@dvogel can you please have a look?

[dvogel]

It looks like this is a case where motif and gtk have different expected char pointers. I'll revert the xpm file changes and cast char* to const char* on the gtk side tomorrow. EDIT: PR here invert the const'ness of the XPM pointer: #14834. I see now why my manual tests didn't catch this. I misread `#ifndef HAVE_XPM` as `#ifdef HAVE_XPM` and mistakenly thought I was compiling the wrong branch.