stack/logrotate: fix copytruncate being accidentally always enabled

In 0409f691 (zope4: Handle log rotation following Zope documentation.,
2022-03-10) we added an option to use copytruncate in logrotate, with
a default value of "false" which is true in jinja2, so it was actually
enabled every where.

This change brings the expected behavior copytruncate option in
logrotate stack, it is not enabled by default, unless explicitly using
copytruncate = true

stack/logrotate/buildout.hash.cfg

@@ -22,4 +22,4 @@ md5sum = 02c1009f8e0dc371cfc1290afef72ec7
 
 [template-logrotate-base]
 filename = instance-logrotate-base.cfg.in
-md5sum = 4e2baa1edd1d27831dda984769102a7c
+md5sum = 303fad78d62d6e29c0c547a9f64fa822

stack/logrotate/instance-logrotate-base.cfg.in

@@ -47,6 +47,8 @@ context =
 # - "post" with commands to execute after rotation
 # - "pre" with commands to execute before rotation
 # - "backup" with directory where to store logs
+# - "copytruncate" to use logrotate's copytruncate option, setting to ""
+#    (the default) disable copytruncate, setting to anything else enable copytruncate
 recipe = slapos.recipe.template:jinja2
 url = {{ logrotate_entry_template }}
 output = ${logrotate-conf-parameter:logrotate-entries}/${:name}
@@ -60,7 +62,7 @@ context =
   key rotate_num :rotate-num
   key nocompress :nocompress
   key delaycompress :delaycompress
-copytruncate = false
+copytruncate =
 post =
 pre =
 frequency = daily