Prevent stream buffer overflow by lowering the rate of incoming partitions

[patkivikram]

Checklist

• [ x] I have included information about relevant versions
• [x ] I have verified that the issue persists when using the master branch of Faust.

Steps to reproduce

Have an agent process slowly. When the number of events on its Queue > stream_buffer_maxsize it causes starvation of other stream processors(agents) processing other topics

Expected behavior

Stop fetching data from the slow topic/partition when the buffer reaches a threshold and then resume it after the pressure drops

Full traceback

Paste the full traceback (if there is any)

Versions

• Python version 3.8
• Faust version 0.3.0
• Operating system centos 7.5
• Kafka version 2.4
• RocksDB version (if applicable) 6.10.3