Summary for NIST bugs

[FX Coudert]

Thanks for the PR numbers I didn't find.

>>* Finally, I have reduced the failure of FM912
>
> Good luck, I didn't understand what they were doing :-)

This is the reduced test case I managed to get (without comment, it's 12 
lines, which I guess is not so bad):
       PROGRAM FM912
       IMPLICIT NONE
       CHARACTER STR*120
       OPEN(FILE='FOO',UNIT=24,ACCESS='DIRECT',RECL=120,FORM='FORMATTED')
C   WRITES RECORDS 1, 2, AND 3
       WRITE(UNIT=24,REC=1,FMT=4126)
4126  FORMAT('AAAAAAAAAAAAAAAA',//, 10HHOLLERITH , T15, 'ONE', 10X,
      1       TL5, 'TWO', TR5, 'THREE', :, 'LAST')
C   READS RECORD 3
       READ(UNIT=24,REC=3,FMT=4127) STR
4127  FORMAT(A120)
       PRINT *, STR
       CLOSE(UNIT=24)
       END

It still exhibits the failure, but I don't really know how to reduce it 
any further. I am scared by the term "HOLLERITH", which has always been 
black magic to me, so I didn't want to work further along this way. 
Perhaps someone will recognise a bug or will be able to reduce it further.

-- 
FX