irnode: When deciding to use an ARR_F for the in-array, check for oparity_dynamic instead of hard-coding End and Sync.

This sort of typeless programming is bad style and deprecated.

- It was only available in the ia32 backend and broken there. As the
  feature is rarely used, I remove it for now instead of going through
  the trouble of fixing it.

The previous approach put the irg link into the block attributes, so
while saving memory an extra indirection was necessary. Some testing
with valgrind massif+callgrind revealed that a direct link increses peak
memory usage by ~1% while also increasing runtime by ~1%. More
importantly the code gets a bit simpler and some strange corner cases can be
removed with the direct link.

This makes our usage of PhiMs to represent the observable behaviour of
non-terminating loops more obvious.
Having an explicit loop Phi constructor also fixes problems with Phi
getting optimized before the keep edge was added.

For details see endless loops section on the homepage.

The semantics on when to set this flag were unclear, so there weren't
any real users anyway.

- We have no_write, pure and terminates now (terminates instead of
  has_loop means that more flags is more specific information, so the
  fixpoint algos are less confusing)
- funccall ana also detects noreturn attribute now (but isn't optimizing
  it)

- Compound type selection is performed with Member
- Array indexing is performed with Sel
- Instead of the simpleSel constructor functions you can use the normal Member
  constructors now
- The concept of array element entities has been removed, Sel nodes just
  contain a pointer to the array type now.

It is unused and anyway redundant due to the given ir_node*.

This can be used to effectively test aliasing in libfirm.

= is_entconst(), is_typeconst, get_entconst_entity() and
get_typeconst_type()