arm, be, ia32, sparc: Do not go on with peephole optimization of IncSP, when be_peephole_IncSP_IncSP() succeeded.

The current IncSP was removed in this case and we now would operate on its predecessor IncSP.
This is wrong, e.g. when asking for currently free registers.
This corrects backend/incsp_pop.c, which was broken on ia32.
It was wrongly assumed that a certain register was free and could be used to turn an IncSP into a Pop.

Otherwise the frontend has to do this after asking the backend for its parameters, which is quite awkward.

Duplicate Projs are a problem when CSE is off, e.g. -O0.

PC relative relocations need %rip as base register so we cannot match
anything else for base.

Calls+Jumps really use PC relative addresses even if they look the same
in the assembler. Create an immediate kind for that, this allows a
slightly simpler bemit_entity().

The following be_peephole_exchange() has a stricter check, i.e. same mode instead of only not mode_T.

It is like new_r_Proj(), but the mode is determined automatically from the out requirement of the predecessor.
This probably resolves some mode mismatches in untested code paths -- AMD64 backend and SPARC hard float, I'm looking at you!

Now register requirements always have have a register class.
This is more consistent and removes some null pointer checks.

Nodes without requirements are never operands of a {Copy,}Keep.

match_requirement() cannot be called with a requirement without a register class.

Now we can use things, which get initialized for the middle end, e.g. modes.

The caller of be_opt_register() calls it, too.