collectd: fix memory stat collection configuration

https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/memory.html#usage-in-bytes says
this is the most efficient approach for v1. It says RSS+CACHE(+SWAP) is the more accurate value.
Elsewhere in the note in https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/memory.html#stat-file,
it says "‘rss + mapped_file” will give you resident set size of cgroup." Overall, it's not clear how
to compute the values so we just use the file.

https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html is better. https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html#memory
says the values are separated out.

src/infra_version.js

@@ -6,7 +6,7 @@
 
 exports = module.exports = {
     // a version change recreates all containers with latest docker config
-    'version': '49.0.0',
+    'version': '49.1.0',
 
     'baseImages': [
         { repo: 'cloudron/base', tag: 'cloudron/base:3.2.0@sha256:ba1d566164a67c266782545ea9809dc611c4152e27686fd14060332dd88263ea' }