The following patch fixes

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87718

   The patch adds a special treatment for moves with a hard register in 
register cost and class calculation.

   The patch was bootstrapped and tested on x86-64 and ppc64.

   I found two testsuite regressions because of the patch.  The expected 
generated code for PR82361 test is too specific.  GCC with the patch 
generates the same quality code but with a different hard register on 
x86-64.  So I just changed the test for  PR82361.

   Another test is for ppc64.  I think the expected generated code for 
this test is wrong.  I'll submit a changed test for a discussion later.

   Although I spent much time on the solution and I think it is the 
right one, the patch is in very sensitive area of RA and may affect 
expected code generation for many targets.  I am ready to work on the 
new regressions as soon as they are found.

   The patch was committed as rev. 260385.