Add max and min functions for mal down sampling

[peachisai]

• Tests(including UT, IT, E2E) are added to verify the new feature.
• Update the CHANGES log.

User Case

metricsRules:
  - name: consumer_latency
    exp: rocketmq_group_get_latency_by_storetime.sum(['cluster','topic','group']).downsampling(MAX)

The opentelemetry metrics

rocketmq_consumer_message_size{cluster="DefaultCluster",broker="broker-a",topic="topic1",group="group2",} 21350.0
rocketmq_consumer_message_size{cluster="DefaultCluster",broker="broker-b",topic="topic1",group="group2",} 7700.0

The sum aggregate method use 'cluster','topic' and 'group' as aggregate-by params,so I can sum the metrics of the two brokers, and then get the max message size instead of the default avg function value

The same scenario for min function

[wu-sheng]

Could you provide the use cases?

[peachisai]

Could you provide the use cases?

I updated the use case

[wu-sheng]

Could you provide the use cases?

I updated the use case

Why do you need a max value? I think in most case, percentile should be better to measure latency? Does max bring a special value?

[peachisai]

Could you provide the use cases?

I updated the use case

Why do you need a max value? I think in most case, percentile should be better to measure latency? Does max bring a special value?

I prefer the maximum value. This looks more intuitive. For example, with size and offset metrics.

[wu-sheng]

Size and offset make sense. But about latency, percentile is better.

[wu-sheng]

Please update the description for the meaningful use cases.

[wu-sheng]

@Entrance is misused here, and this error exists in many other MAL functions. This annotation is used in MAL engine use, in MAL, accept method is the main entrance by default.

[wu-sheng]

I am trying to fix existing codes.

[peachisai]

Please update the description for the meaningful use cases.

Updated