Inlining, nested functions, and VLAs

[Richard Kenner]

    I suppose we could force a type match by adding a VIEW_CONVERT_EXPR...

We could.  It would be the very safe change of changing the abort to
a build of a VIEW_CONVERT_EXPR.  However, the validity check that's there
is valuable, I think, and I'm not sure that it'd be worth it to remove
that check just for this obscure case.  On the other hand, it's certainly
far less code than adding a check for and disallowing this inlining.