largefiles: don't actually remove largefiles in an addremove dry run
The addlargefiles() method already properly handled dry runs.
$ hg init repo
$ cd repo
$ touch foo
$ hg ci -Am 'add foo'
adding foo
$ hg up -C null
0 files updated, 0 files merged, 1 files removed, 0 files unresolved
this should be stored as a delta against rev 0
$ echo foo bar baz > foo
$ hg ci -Am 'add foo again'
adding foo
created new head
$ hg debugindex foo
rev offset length ..... linkrev nodeid p1 p2 (re)
0 0 0 ..... 0 b80de5d13875 000000000000 000000000000 (re)
1 0 24 ..... 1 0376abec49b8 000000000000 000000000000 (re)
$ cd ..