Add metadata to segment tracking

[ChristopheDuong]

What

I'm proposing to add some properties to segment tracking.

If I look at data sent by Segment, for example:

select *
FROM `ab-analytics-308500.api.connector_jobs` 
where job_id = 853 and connection_id = '4997a08c-5f27-4d1b-afd0-427177c0ea8d'
order by timestamp

I see that some tracking data for a job was sent on multiple different dates (all dates columns are auto-generated and handled by Segment)

see results:
query_results.txt

(Looking at the full source dataset from segment, I find 6 different dates for the same ids all the way back from 2021-09-22)

How

• Airbyte should attach some date data to the metadata payload of when the tracking call is triggered, so we don't rely on the common fields handled by Segment: https://segment.com/docs/connections/spec/common/#timestamps
• Additionally, some docker repositories may be masked and overridden (for example in the case of SSL-strict destinations). So the connector definition id is not enough to distinguish between the actual connector implementation. Surfacing the docker repository would help to properly track this.

[sherifnada]

should we also add end time?

[ChristopheDuong]

duration is computed using the end time and we now have start time so we can always re-compute the reverse.
Let's leave it out to make the amount of data transiting through segment a little smaller