I had 2 databases and 2 shards - ycsb_test database on shard rs0 and ycsb_test2 on shard rs1 (collections were NOT sharded).
I then tried to remove the shard rs1 and what happened is:
1. draining of chunks from collection config.system.sessions begin and finished (since this collection is a system collection and is always sharded)
2. operator just got into loop putting into logs: "draining ongoing","shard":"rs1","chunk remaining":0
since the chunks collection migration finished, but it still cannot remove the shard because there is one unsharded collection there.

Here's how it looks:
Operator logs:

sharding status from mongos:

And here's the reason why it will not end:

So we need to drop the ycsb_test2 database OR use movePrimary to make the mongo migrate the database to another shard.

https://docs.mongodb.com/manual/reference/command/movePrimary/

Notes:
1. it's not clear from operator logs why it is in the loop
2. for these databases/collection should we execute movePrimary via operator or just make a user do that, but in that case make it clear in the logs that it needs to be done
3. if it will need to be done manually, we should add this into documentation

After I executed

data was moved to shard rs0, shard rs1 removed and I got message in operator logs: