Add optional Cluster instance limit

Right now we do not limit the number of Cluster instances created, but in order to avoid misuse we could add a feature a user configure. Something like:

• On Cluster, keep a static counter of instances (decremented on shutdown)
• By default there is no upper limit, but we could query it for debugging reasons (the counter)
• Optionally allow to set a static limit and if it is reached prevent from creating new instances (i.e. in cluster#connect)

Maybe this can be put into Core instead so all SDKs could benefit from it, but for usability those static setters should still be exposed on the Cluster.

Related: is there something similar we should do for the environment?