varobj's find_frame_addr_in_frame_chain hack (and other ARM unwind notes)

[Andrew Cagney]

> varobj.c:find_frame_addr_in_frame_chain, and mi-var-display.exp.  We print
> out $fp, and use then search for that value in the frame chain.  Well,
> the DWARF-2 CFA value on ARM is biased from the frame pointer register,
> generally by four bytes.  I chose to use the same thing in the prologue
> unwinder for consistency - it makes unwinding ARM_SP_REGNUM simpler, which
> we need to get right in order for dummy frames to work.

Two wrongs, and they don't make a right :-(

- the arm shouldn't be setting / forcing FP_REGNUM
Instead, "fp" should always be the frame base.
- the test shouldn't be using $fp to re-find a frame
It should be using a frame ID, but that involves interface changes to 
the MI -> MI 4.

Andrew