gfortran patch for PR28974 Slow compilation of data statements.

[Paul Thomas]

Jerry,
>>
>>> While working pr24978, 
A bug that is related to pr24978 is pr29786 - in this case it is 
associated with data values that do NOT overlap but gfc insists that 
they do because they are equivalenced.  I have tried picking around at 
the layout of the union in trans-common.c but kept running into refusals 
to play by the backend.  The backend will not let me or, more 
accurately, I do not know how to make a constructor, when the data 
elements are intermeshed.

I think that the path of least resistance is to do what a man has to do 
in the frontend; eg. to resolve the equivalence and to transfer the data 
statement to the equivalence element  that.... well what exactly?  
trans-common.c tries to layout the intitialization in order of offset, 
so vacuuming up and transferring data statements has to occur in the 
same way.

Cheers

Paul