lfdiff.c redefines global "cksum" (One-Line fix)

[voretaq7]

Pretty much just the subject: lfdiff.c (re)defines the global variable cksum even though lfdiff gets linked with transcript.o (which already defines this).
The CLANG shipped with FreeBSD 13 is now "particular" about this when linking & throws an error.

It looks like a trivial one-line fix - delete the definition from lfdiff.c line 47 - so not really worth a PR.
(Testing this internally to be certain but it seems like the obvious and correct answer.)

[sth0]

Looks to me like almost all the utilities do this. Is lfdiff different in some way? Should they all have an "extern" tag? Such as retr.c, stor.c, and transcript.h. Since most of these utilities include transcript.h should they just all remove the re-definition?

% grep -n '^int.*cksum' *.[ch]
daemon.c:54:int cksum = 0;
ktcheck.c:62:int cksum = 0;
lapply.c:48:int cksum = 0;
lcksum.c:33:int cksum = 0;
lcreate.c:58:int cksum = 0;
lfdiff.c:47:int cksum = 0;
lmerge.c:24:int cksum = 1;
t2pkg.c:43:int cksum = 0;
transcript.c:54:int cksum;

% grep -n '^extern.*int.*cksum' *.[ch]
retr.c:50:extern int cksum;
stor.c:49:extern int cksum;
transcript.h:37:extern int cksum;

[hfath]

Wonderful... what is the significance of the lmerge.c:24:int cksum = 1;?

Otherwise, I'd say (and will try out at some point) to remove all re-definitions, provided the files all #include "transcript.h".