Slack: user groups speedup

[kalisp]

Brief description

Query user and group information only if necessary.

Description

When message contains '@' character, translation between users and Slack user id is necessar. This operation might hit rate limit and slows down publish process.
Trigger this querying only if absolutely necessary.

Testing notes:

1. publish something that should trigger Slack notification
2. compare messages with and without @ character

[BigRoy]

What's the speed of client.get_users_and_groups()? Is it very slow?

Since this is an InstancePlugin would it be maybe worth caching the result in context.data on the first run and any consecutive instances would re-use the value.

For e.g. 10 instances that'd bring it back from 10 calls to a single call.

Kind of like this:

OpenPype/openpype/hosts/maya/plugins/publish/collect_inputs.py

Lines 110 to 126 in 4a4bb22

	def process(self, instance):
	# For large scenes the querying of "host.ls()" can be relatively slow
	# e.g. up to a second. Many instances calling it easily slows this
	# down. As such, we cache it so we trigger it only once.
	# todo: Instead of hidden cache make "CollectContainers" plug-in
	cache_key = "__cache_containers"
	scene_containers = instance.context.data.get(cache_key, None)
	if scene_containers is None:
	# Query the scenes' containers if there's no cache yet
	host = registered_host()
	scene_containers = list(host.ls())
	for container in scene_containers:
	# Embed the members into the container dictionary
	container_members = set(get_container_members(container))
	container["_members"] = container_members
	instance.context.data["__cache_containers"] = scene_containers

[BigRoy]

Suggested change

	cache_key = "__cache_ids"
	cache_key = "__cache_ids"

Note that the context data is accessible by all plug-ins. It might be useful to avoid clashing with anything else that'll try to cache ids or alike like this.

Would: self.__class__.__name__ + "__cache_ids" be a good candidate? or at least "__slack_user_cache"?