[feat]: implement monitoring for each (multi-sink) node

[Kristina-Pianykh]

What needs to be monitored:

1. rate of the partitioning input i (cumulative; compute for all multi-sink nodes)
2. rates of the other inputs to the multi-sink query Q (the value is non-cumulative)
3. rate of Q (cumulative; compute for all multi-sink nodes)

What needs to be pre-computed by the launch of the program on each node:

• determine the multi-sink query Q
• how many nodes process Q
• the partitioning input i
• const value of the stainer tree

[Kristina-Pianykh]

Implementation details here: https://github.com/Kristina-Pianykh/flink-multimetric-alert

Next step: incorporate into the main program

[Kristina-Pianykh]

Done in https://github.com/Kristina-Pianykh/flink-multinode/commit/e11b4b94e26c08c5c9fb7acbae316a0916eaf508 and https://github.com/Kristina-Pianykh/flink-multinode/commit/977b942d28e20b9dac0e7056e7c415e749984808

[Kristina-Pianykh]

EDIT: rewrote in https://github.com/Kristina-Pianykh/flink-multinode/pull/7