On page load, we load ALL channels and channel members from all teams.
But then, on team_switch, we would again load channels and channel
members from that team. This was redundant and mainly kept
because previously the websocket events were considered unreliable.
Now with reliable websockets, and client-side pings, we can detect
broken connections faster and recover without loss.
Additionally, the getAllChannelMembers call would page through
all responses on the client side. This was inefficient and incur
extra latency. To optimize for this, we introduce server-side
streaming of the full response if page is set to -1.
This optimizes the intial response as well.
https://mattermost.atlassian.net/browse/MM-56906
```release-note
Optimize team switch operation by removing calls to get channels
and channel members.
```
Co-authored-by: Mattermost Build <build@mattermost.com>
* Add ChannelStore.UpdateMultipleMembersNotifyProps
* Make UpdateMultipleMembersNotifyProps return updated values from the DB
* Add UpdateChannelMembersNotifications plugin API
* Extract i18n
* Fix style
* Make layers
* Change to PatchMultipleMembersNotifyProps
* Add limit to PatchChannelMembersNotifyProps
* Add additional unit tests
* Address feedback
* Lowercase decodeJSON
* Have PatchMultipleMembersNotifyProps update LastUpdateAt
* Fix tests that relied on unreliable return order
* Fix i18n
