Possible race condition when resetting a subscription to head

[matt2e]

Was noticed during PR review, but not sure how important it is:
#2210

Possible case:

• Subscription is behind and currently consuming an event
• Reset subscription command comes in sets the state to idle and updates the cursor to head
• 2 new events are posted to the topic
• New event starts being consumed by the subscription
• Original event (from before the reset command) finishes being processed, sets the state to idle even though we are processing the other event still
• Then pub sub triggers the next event to be processed (without knowing the other event is being processed still)

[matt2e]

From original comment by @worstell:

ah interesting, do you think the solution would be to only update the cursor and not the state? so that if there are any outstanding async calls executing, they'd have to complete before any new ones are kicked off

Honestly I'm not sure. Maybe that would be the neatest solution?