Fix committed offset is always less than log-end-offset by 1

[shamanu4]

Description

This PR fixes a bug, that committed offset is always less than log-end-offset in Kafka by 1 when all messages were processed. This bug causes 2 problems:

• after the worker restart, the last message from topic partition was re-send to the consumer. (Was fixed by this check in the _drain_messages function: if committed is None or offset > committed. )
• if there was an offset reset in the partition to the latest, consumer lag is equal to 0. The next message that appears in this partition will be ignored by the consumer due to the previous check, in the _drain_messages function.

This PR increases any committing offset by 1, so the log-end-offset will be equal to 0 when all messages are read.

Fixes #620

[shamanu4]

Lint errors seem to be unrelated to this PR.

[rochapaulo]

I am facing this issue, need this fix!

[CLAassistant]

All committers have signed the CLA.

[sivasai-quartic]

I'm also facing the same issue, need this fix

[patkivikram]

Can you create this fix on the faust-streaming project? We can merge this to master and create a new release

[leah4kosh]

hey, any updates on that? we need it :-)

[patkivikram]

This is already available in faust-streaming