Load balancing non existent? #15
Description
As far as I can see when running on a single function app instance it will only currently process changes from 4 physical partitions.
In the core change feed processor 0 is used as maxPartitionCount in the EqualPartitionsBalancingStrategy so there is no such limitation imposed.
There also seems to be no logic to redistribute processing if a function app scales in and the number of hosts decrease (as lease allocation is only called at startup). In the core API project this is controlled by the LeaseAcquireInterval to fairly frequently check if any rebalancing is needed.
I do wonder about the wisdom of creating a new project for this vs getting it integrated into the core project so it can benefit from all the work already done there (seems like it should be possible to come up with common abstractions that both APIs can use)