fix(ingest/mongodb): support disabling schemaSamplingSize

[diegoreico]

fix: allow mongodb source connector to have a 0 input for field schemaSamplingSize

Fixes #9287

Checklist

• The PR conforms to DataHub's Contributing Guideline (particularly Commit Message Format)
• Links to related issues (if applicable)
• Tests for the changes have been added/updated (if applicable)
• Docs related to the changes have been added/updated (if applicable). If a new feature has been added a Usage Guide has been added for the same.
• For any breaking change/potential downtime/deprecation/big changes an entry has been made in Updating DataHub

[hsheth2]

It seems like the current behavior will already do the right thing if schemaSamplingSize is set to null. I might be missing something, but it seems like it'd make more sense to simply fix the docs here and leave the actual code unchanged

Suggested change

	description="Number of documents to use when inferring schema size. If set to `0`, all documents will be scanned.",
	description="Number of documents to use when inferring schema size. If set to `null`, all documents will be scanned.",

[diegoreico]

I've made a small update on the linked issue on my first message

As far as I have reached, I think that it's imposible to force a full mongodb collection scan, because the yaml parser + validator forces to provide a value greater than 0, but internal code checks for null

I'm gonna try to explain my thoughts on detail:

1. schemaSamplingSize: Optional[PositiveInt] does allow null value, but not 0 value so I changed it to NonNegativeInt
2. schemaSamplingSize is also asigned to a Pydantic Field with the following default -> default=1000,. So, if no value is provided, this will have that default value
3. You can't explicit provide a null or none value trough the yaml parser (as i also reflect now on the issue)

So, if I need to provide a value to the parser and it can't be null, providing a 0 value could be a good option.

[hsheth2]

@diegoreico I've made a small tweak to your code - let me know what you think

Setting schemaSamplingSize: 0 will be invalid, but you should now be able to explicitly set schemaSamplingSize: null and have that work properly.

We actually had both in our documentation, so my hope is this makes it consistent and avoids having a "magic value" of 0.