feat/add histogram aggregator in defaultAggregator

[NeJan2020]

Because we have started to use a stand-alone aggregator instead of OTLP-SDK, some functions that were originally implemented by the OTLP-SDK aggregator are gradually being moved to the stand-alone aggregator. Histogram aggregation is one of them and will be needed in standard version of kindling

During this process, I ran into some problems and decided to make some modifications to the model.Gauge structure

Reasons for changing the model of Gauge

1. If the aggregator needs to support Histogram output, the ExplicitBoundaries and BucketCounts information must be included in the output. The current structure is not suitable for storing this kind of information.
2. This modification has little damage to the structure and is less difficult to modify, but it improves the scalability of the structure.

Reasons for keeping the model of Gauge

1. As the underlying data structure of the pipeline, Gauge is used too widely. Many components need to be adjusted
2. The original Gauge structure has simple functions and is easy to understand. Not all developers have the basic knowledge of Histogram type data.

Because implementing the Histogram based on modifying the Gauge structure is far simpler than not modifying it. I provided this PR which provides a simple implementation to illustrate the impact of Gauge changes on other components. At the same time, it has been able to support the aggregation of Histogram type data.

Description

1. BREAKCHANGE: Change the model of Gauge
2. Add a new aggregator value named histogram in the default aggregator

Motivation and Context

The existing performance of the aggregator based on the OTLP-SDK cannot support our needs, and our new aggregate processor can not support Histogram.

How Has This Been Tested?

cd collector && go test -v github.com/Kindling-project/kindling/collector/pkg/aggregator/defaultaggregator
=== RUN Test_aggValues_sum
--- PASS: Test_aggValues_sum (0.00s)
=== RUN Test_aggValues_sum_histogram
--- PASS: Test_aggValues_sum_histogram (0.00s)
=== RUN Test_aggValues_max
--- PASS: Test_aggValues_max (0.01s)
=== RUN Test_aggValues_max_histogram
--- PASS: Test_aggValues_max_histogram (0.00s)
=== RUN Test_aggValues_avg
--- PASS: Test_aggValues_avg (0.01s)
=== RUN Test_aggValues_avg_histogram
--- PASS: Test_aggValues_avg_histogram (0.01s)
=== RUN Test_aggValues_last
--- PASS: Test_aggValues_last (0.00s)
=== RUN Test_aggValues_last_histogram
--- PASS: Test_aggValues_last_histogram (0.00s)
=== RUN Test_aggValues_count
--- PASS: Test_aggValues_count (0.00s)
=== RUN Test_aggValues_count_histogram
--- PASS: Test_aggValues_count_histogram (0.00s)
=== RUN Test_aggValues_histogram
--- PASS: Test_aggValues_histogram (0.01s)
=== RUN Test_aggValues_histogram_histogram
--- PASS: Test_aggValues_histogram_histogram (0.01s)
=== RUN Test_defaultValuesMap_sum
--- PASS: Test_defaultValuesMap_sum (0.00s)
=== RUN Test_defaultValuesMap_avg
--- PASS: Test_defaultValuesMap_avg (0.00s)
=== RUN Test_defaultValuesMap_max
--- PASS: Test_defaultValuesMap_max (0.00s)
=== RUN Test_defaultValuesMap_lastValue
--- PASS: Test_defaultValuesMap_lastValue (0.00s)
=== RUN Test_defaultValuesMap_countValue
--- PASS: Test_defaultValuesMap_countValue (0.00s)
=== RUN Test_defaultValuesMap_histogramValue
--- PASS: Test_defaultValuesMap_histogramValue (0.00s)
=== RUN TestConcurrentAggregator
--- PASS: TestConcurrentAggregator (0.56s)
=== RUN TestRecord
--- PASS: TestRecord (0.00s)
PASS

[NeJan2020]

The raw struct of Gauge is

type Gauge struct {
	Name  string
	Value int64
}

The new struct of Gauge is

type Gauge struct {
	Name string
	//	Data can be assigned by:
	//	Int
	//	Histogram
	Data isGaugeData
}

type Int struct {
	Value int64
}

type Histogram struct {
	Sum                int64
	Count              uint64
	ExplicitBoundaries []int64
	BucketCounts       []uint64
}

type isGaugeData interface {
	isGaugeData()
}

func (*Gauge_Int) isGaugeData()       {}
func (*Gauge_Histogram) isGaugeData() {}

type Gauge_Int struct {
	Int *Int
}

type Gauge_Histogram struct {
	Histogram *Histogram
}

The implementation details of this structure refer to the oneof implementation of the protocol in golang

[NeJan2020]

Here I changed the self-telemetry metric, will this influence other components?

[dxsup]

No.