Fix portability problems recently exposed by regression tests on Alphas.

1. Distinguish cases where a Datum representing a tuple datatype is an OID
from cases where it is a pointer to TupleTableSlot, and make sure we use
the right typlen in each case.
2. Make fetchatt() and related code support 8-byte by-value datatypes on
machines where Datum is 8 bytes.  Centralize knowledge of the available
by-value datatype sizes in two macros in tupmacs.h, so that this will be
easier if we ever have to do it again.

src/include/access/htup.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: htup.h,v 1.41 2000/11/30 08:46:25 vadim Exp $
+ * $Id: htup.h,v 1.42 2000/12/27 23:59:13 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -147,6 +147,9 @@ typedef struct xl_heap_update
 #define MaxAttrSize		(10 * 1024 * 1024)
 
 
+/*
+ * Attribute numbers for the system-defined attributes
+ */
 #define SelfItemPointerAttributeNumber			(-1)
 #define ObjectIdAttributeNumber					(-2)
 #define MinTransactionIdAttributeNumber			(-3)
@@ -156,9 +159,6 @@ typedef struct xl_heap_update
 #define TableOidAttributeNumber			        (-7)
 #define FirstLowInvalidHeapAttributeNumber		(-8)
 
-/* If you make any changes above, the order of offsets in this must change */
-extern long heap_sysoffset[];
-
 /*
  * This new HeapTuple for version >= 6.5 and this is why it was changed:
  *