source-mongo: Warn (vs fail) on different _id types in collection

[evantahler]

Re: https://github.com/airbytehq/oncall/issues/5379
Closes: https://github.com/airbytehq/oncall/issues/5379
Dev Image: https://github.com/airbytehq/airbyte/actions/runs/9393312995

When we encounter mongo collections with multiple _id types in the same collection, ordering the records for the initial snapshot gets weird. We can't fully guarantee that _id:1 is always sorted before or after _id:"b". However, failing the sync because of these issues, in all cases, is also premature until we can confirm data loss more robustly. In the linked OC issue, we do have signal that at least for some customers, when syncing with multiple ID types, we do not loose data.

This PR changes a throw into a logger.warn for now.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
airbyte-docs	⬜️ Ignored (Inspect)	Visit Preview		Jul 22, 2024 10:55pm

[rodireich]

@evantahler I think this is probably the way to go - a single misbehaving row shouldn't cause an entire sync to fail.
In addition to removal of the strict test,
What I think we can also add here to formalize it is:

1. We don't want a single row to trump a sync, but some threshold - if id's are all over the place then how we determine the "correct" type.
2. Marking incorrect id type rows with a meta event - similarly to how we do on other connectors when we fail to convert a value.
3. Add analytics to be able to know how common this is

[evantahler]

We have learned that ins some cases, it is intentional that there are many types of _ids in a single collection, so it's more than a single bad object :(

[theyueli]

Thanks for this change!