feat(sdk-metrics-base): detect resets on async metrics

[legendecas]

Which problem is this PR solving?

Detect resets on async metrics.

Fixes #2987

Short description of the changes

• Add monotonicity support in SumAggregator
  • If the SumAggregator is monotonic, it performs SumAggregator.diff and returns the current value if the previous value is greater than the current one.
  • Negative values are ignored in monotonic SumAccumulation.
  • SumAggregator.merge distinguishes a reset delta, and does not merge it with previous accumulation.
• Added reset and gaps detection for async metric instruments.
  • CUMULATIVE reader now collected the reset value and startTime.
  • Fixed a problem that DELTA reader may receive negative values on resetting a monotonic metric.
• Resets on async metrics also resets the data points' exported start time.
• The startTime and endTime now follow the spec definitions (reset & gaps):
  • For points with delta aggregation temporality, the startTime of each point matches the endTime of the preceding point (for that particular reader).
  • Otherwise, the startTime of each point matches the startTime of the initial observation.

For an event stream, we should provide the following metric exports:

Async Monotonic Event Inputs	Cumulative Exports	Delta Exports
1, Event Time T1	1, StartTime T1, EndTime T1	1, T1-T1
100, T2	100, T1-T2	99, T1-T2
200, T3	200, T1-T3	100, T2-T3
1, T4 (Reset)	1, T4-T4	1, T4-T4
200, T5	200, T4-T5	199, T4-T5
(A gap, no inputs), T6	200, T4-T6	(no output)
300, T7	300, T4-T7	100, T7-T7

Async Non-monotonic Event Inputs	Cumulative Exports	Delta Exports
1, Event Time T1	1, StartTime T1, EndTime T1	1, T1-T1
100, T2	100, T1-T2	99, T1-T2
200, T3	200, T1-T3	100, T2-T3
1, T4	1, T1-T4	-199, T3-T4
200, T5	200, T1-T5	199, T4-T5
(A gap, no inputs), T6	200, T1-T6	(no output)
300, T7	300, T1-T7	100, T7-T7

For a sync event stream, the exported times are:

Sync Event Inputs	Cumulative Exports	Delta Exports
100, Event Time T1	(not exporting)	(not exporting)
(No event), T2	100, T1-T2	100, T1-T2
100, T3	(not exporting)	(not exporting)
100, T4	(not exporting)	(not exporting)
(No event), T5	300, T1-T5	200, T2-T5
(No event), T6	300, T1-T6	(no output)

Type of change

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

• Monotonic SumAggregator and Non-monotonic SumAggregator
• Add checks on SyncMetricStorage exported startTime and endTime
• Add checks on AsyncMetricStorage exported startTime and endTime

Checklist:

• Followed the style guidelines of this project
• Unit tests have been added

[codecov]

Codecov Report

Merging #2990 (40d19c3) into main (4378303) will decrease coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #2990      +/-   ##
==========================================
- Coverage   93.10%   93.08%   -0.03%     
==========================================
  Files         188      188              
  Lines        6224     6261      +37     
  Branches     1316     1323       +7     
==========================================
+ Hits         5795     5828      +33     
- Misses        429      433       +4     

Impacted Files	Coverage Δ
...ntelemetry-sdk-metrics-base/src/aggregator/Drop.ts	100.00% <ø> (ø)
...telemetry-sdk-metrics-base/src/aggregator/types.ts	100.00% <ø> (ø)
...metrics-base/src/state/MeterProviderSharedState.ts	100.00% <ø> (ø)
...emetry-sdk-metrics-base/src/state/MetricStorage.ts	100.00% <ø> (ø)
.../opentelemetry-sdk-metrics-base/src/Instruments.ts	94.73% <100.00%> (+0.14%)	⬆️
...metry-sdk-metrics-base/src/aggregator/Histogram.ts	91.04% <100.00%> (-5.84%)	⬇️
...metry-sdk-metrics-base/src/aggregator/LastValue.ts	100.00% <100.00%> (ø)
...entelemetry-sdk-metrics-base/src/aggregator/Sum.ts	100.00% <100.00%> (ø)
...y-sdk-metrics-base/src/state/AsyncMetricStorage.ts	100.00% <100.00%> (ø)
...sdk-metrics-base/src/state/DeltaMetricProcessor.ts	100.00% <100.00%> (ø)
... and 8 more

[dyladan]

Sorry for late review. Large PR is hard to get time to really go through. Most changes are just passing the hrtime though. I added some comments but none of them are blocking.

[pichlermarc]

Finally had the time to look through this. 🙂
Just a few non-blocking (mostly docs related) comments. Thanks for taking the time to work on this! 🙂

[legendecas]

Thank you for taking a look at this!