[committed] allow assert in inner class

[Per Bothner]

Andrew Haley wrote:
> Thanks.  3.4 too?  Or is this not sufficiently important?

I don't know.  I don't think it fixes a regression, but it
does close two PRs.

I'd lean towards including it in 3.4; it seems fairly safe.
-- 
	--Per Bothner
per@bothner.com   http://per.bothner.com/