[rfa] mips heuristic_proc_start fix

[Andrew Cagney]

> 
> Nope!  We just read pc off the stack.  If the PC is legitimate in the
> first place, none of these nasty loops are a problem.  My pc was 0x2;
> that's bad, but not cause for gdb to die.  What'd you say to a warning
> and early return?


(Technically, before the call to ADDR_BITS_REMOVE() both 3 and 1 are 
valid, after 2 is valid - MIPS16 code.  :-)

I think just having the one message is easier.  As the comment notes:

                 /* This actually happens frequently in embedded
                    development, when you first connect to a board
                    and your stack pointer and pc are nowhere in
                    particular.  This message needs to give people
                    in that situation enough information to
                    determine that it's no big deal.  */

so any other message would need to be just as long.

	Andrew