[patch+7.4] reread.exp 7.3->7.4 regression

[Joel Brobecker]

> The freeing is not the only problem.  The same problem applies on the lines
> accessing the elements, those statements can also crash now without this
> patch.  It is just not shown in this crash backtrace.

Ugh, this is really nasty. I know you've delved into this more than
I have, so I am Ok with your proposed initial work around. Others
might have an opinion too. Is it worth labeling all the places that
we ended up modifying? I guess not - they all reference the new
field which is clearly documented as being here to work around
a deficiency in the design.

I think this problem ties us with the problem Tom was trying to
figure out, where BFD opens and closes bfd's without telling us.

-- 
Joel