If we can prove that an entity is never written to, replace all loads to
the entity by the value of its initializer.

Some optimizations, e.g., inlining, can create tuples with tuples as
operands.  The current code in determine_entity_usage cannot deal with
this situation and computes wrong usage information.  For now, we just
remove tuples before analysis.

Printing a large double (i.e., 1e256) exceeded the buffer size.

This save quite a bit of code duplication.
Also the code drifted apart in the different backends, which this macro prevents now.

* IR_GRAPH_PROPERTY_NO_TUPLES is cleared automatically by the prior call to turn_into_tuple().
* The function does not create any Bads, so clearing IR_GRAPH_PROPERTY_NO_BADS is not necessary.

Now the block scheduler can remove many empty blocks.

be: Remove the unused function 'be_get_type_unsigned_long_long()' as well as 'type_unsigned_long_long' from 'struct backend_params'.

be: Remove the unused function 'be_get_type_long_long()' as well as 'type_long_long' from 'struct backend_params'.

'movq' is not accepted by older assemblers, but 'movd' works for both older and newer ones.
objdump always disassembles to 'movd' anyway.

The number is small and never a bitmask, so decimal is more readable in this case.

This avoids code duplication and actually works correctly.
This fixes opt/convs.c.

Computing (post)dominators removes tuples, which in turn invalidates out edges, if any tuples were removed.

It currently uses dominance checks, though the use looks questionable.
backend/lower_call4.c is the only test case which triggers this check.
The check was added in bd1b1f0e.
Though later in 5f4cf6e0 another check was added, which would also handle this case for different reasons (target is not on the frame).
So at least the test case needs to be updated validate the dominance check.
Maybe the dominance check should be replaced by a dependence check.

This should fix opt/jumpthreading4.c and opt/place.c when debug (in particular verification) is off.