DRIVERS-3036 Remove the BSON document size validation requirement for the client bulk write operation

[stIncMale]

Preview

Please complete the following before merging:

• Update changelog.
• Test changes in at least one language driver.
• Test these changes against all server versions and topologies (including standalone, replica set, sharded
  clusters, and serverless).

DRIVERS-3036

[stIncMale]

@jyemin I dislike how the change turned out - the rules became complex for no good reason. I think the spec should stay unchanged, even if its requirement does not match Java's behavior for the rest of commands, especially given that the current (before the PR) spec requirement follows the "Where possible, depend on server to return errors" principle.

[stIncMale]

Does anyone understand why the pre-commit failed? https://github.com/mongodb/specifications/actions/runs/11696304981/job/32573170376?pr=1725

The only error-related info I see there is

Error: Process completed with exit code 1.

and it's not helpful.

[alcaeus]

Does anyone understand why the pre-commit failed?

Opening up the details of the pre-commit task shows the diff it produced: https://github.com/mongodb/specifications/actions/runs/11713782255/job/32627198204?pr=1725#step:4:57

diff --git a/source/crud/bulk-write.md b/source/crud/bulk-write.md
index 569e9a2..8724232 100644
--- a/source/crud/bulk-write.md
+++ b/source/crud/bulk-write.md
@@ -912,8 +912,8 @@ In accordance with the
 ["_Where possible, depend on server to return errors_"](https://github.com/mongodb/specifications/blob/f8dbd2469f18d093f917efa1f758024bca5d3aaa/source/driver-mantras.md#where-possible-depend-on-server-to-return-errors)
 mantra, drivers should rely on the server validating the size. Such reliance is not possible when a server
 acknowledgement is not expected, but for simplicity and given that unacknowledged writes does not provide a way to know
-whether a write succeeded on the server, it does not seem necessary to help an application with that info
-in the specific situation when the BSON document size limit is exceeded.  
+whether a write succeeded on the server, it does not seem necessary to help an application with that info in the
+specific situation when the BSON document size limit is exceeded.
 
 ## **Changelog**

@GromNaN since you already added logic to create a commit for updated JSON test files, do you think we can somehow also apply suggestions from the linter? I suppose it's not as easy if both jobs need to make changes, as the commits may happen near-simultaneously and conflict when pushed.

[stIncMale]

Thanks, @alcaeus, for pointing out what exactly was wrong. I initially failed to interpret the logs correctly.

[kevinAlbs]

Thank you very much for the PR. LGTM with minor suggestions.

---

Does anyone understand why the pre-commit failed?

README > Linting describes how to run locally. Running the following works for me:

pre-commit run --all-files --hook-stage manual mdformat