Fix server crashed, regular cleanups, improve configs, etc...

[hweihwang]

Issue

The whiteboard server crashes after running for 25+ minutes (#258) due to:

• The socket data gets invalidated/removed but the socket is still active
  => Mostly because of the LRU cache evicting data while the socket connection is still alive

new LRUCache({
max: 1000,
ttl: 30 * 60 * 1000,
ttlAutopurge: true,

(My mistake for not separate the storage of Rooms Data and socket/session data)

Fix

• Using normal in memory storage for storing socket data (session data) instead of LRU (still along with redis for cluster setups) since the socket data can live as long as the connection alive and we can manually manage that easily (unlike Rooms data a bit more tricky when manually manage)
• Better try catch to handle exception
• Doing some regular cleanups for those LRU( Or In-mem) data (Feature Request: configure automatic saving #295 can be involved since we'll do the savings along with too)

[juliusknorr]

I know it is in draft and I haven't checked the code yet, but maybe you can provide a short summary on why it crashed and how the fix changes that to the PR description :)

[hweihwang]

Picking configs, constants improvements from https://github.com/nextcloud/whiteboard/pull/226/files for not blocking merge

[juliusknorr]

Small comment, otherwise that looks good 👍