feat(reduce transform): add max_groups option to bound memory usage

[karan-vk]

Summary

Adds a max_groups option to the reduce transform that limits the number of groups kept in memory at any one time. This is a safeguard against unbounded memory growth (and eventual OOM) when group_by contains high-cardinality fields.

When a new group would exceed the limit, the group that has gone the longest without receiving an event is flushed downstream to make room, so no data is dropped. Each early flush increments the new reduce_max_groups_exceeded_total internal metric and logs a rate-limited warning.

When max_groups is not set, behavior is unchanged.

Vector configuration

transforms:
  my_reduce:
    type: reduce
    inputs: ["my_source"]
    group_by: ["request_id"]
    max_groups: 10000

How did you test this PR?

• Added unit tests: max_groups: 0 is rejected at config parse time, and when the limit is exceeded the least recently updated group is flushed while events for existing groups do not trigger a flush.
• cargo test --lib transforms::reduce (17 passed)
• cargo clippy --lib --workspace -- -D warnings
• cargo fmt --all -- --check
• cargo vdev check events
• Regenerated component docs with cargo vdev build component-docs

Is this a breaking change?

• Yes
• No

Does this PR include user facing changes?

• Yes. Please add a changelog fragment based on our guidelines.
• No. A maintainer will apply the no-changelog label to this PR.

References

• Closes: Safeguard reduce transform memory use #3027
• Related: Add configuration for limiting reduce transform #9498

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[datadog-vectordotdev]

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: b3f84a0 | Docs | Datadog PR Page | Give us feedback!}

[karan-vk]

I have read the CLA Document and I hereby sign the CLA