[processor/deltatocumulative]: observe accumulation metrics

[sh0rez]

Description:

Uses the otel meter to observe some key metrics about the accumulation:

Name	Description
streams_count	Number of streams currently tracked by the aggregation state
datapoints_processed	Total number of datapoints processed, whether successful or not
datapoints_dropped	Faulty datapoints that were dropped due to the reason given in the reason attribute
seconds_lost	Total length of all gaps in the streams, which occur e.g. due to lost in transit

Link to tracking Issue:
#30705

Testing: None

Documentation: Readme updated

[ethercrow]

When I was trying out this processor I was wondering how to debug it and ended up pretty much just reading all the code and adding Printfs. The set of metrics from this PR would have helped a lot!

One nitpick on naming: samples is a very overloaded word and at least my first thought was that it's about sampling. Maybe accepted_datapoints and dropped_datadoints would be clearer?

[ethercrow]

I don't understand the need for the addable interface. My best guess is that it takes 12 lines to make 4 other lines slightly shorter. Is it really all it's doing?

Otherwise looks fine to me.

[sh0rez]

@RichieSams @djaglowski I think this is ready for final review, please take another look!

[jpkrohling]

The main code looks nice, but I'm missing tests for the new behavior that was introduced.