Add memory resource parameter to cudf::mark_join constructors

[PointKernel]

Description

Follow-up to #23264 and #23277.

This PR adds an mr parameter to the cudf::mark_join constructors to control allocations for the persistent cuco hash table and optional bloom-filter prefilter, aligning the wrapper with the temporary MR effort (#20780).

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[PointKernel]

/ok to test 0ae3e47

[coderabbitai]

📝 Walkthrough

Summary by CodeRabbit

• New Features
  • Mark-based joins now support an optional custom device memory resource.
  • Existing behavior remains unchanged by default, using the current device resource automatically.
• Bug Fixes
  • Improved memory allocation control for join processing and optional bloom-filter prefiltering.
• Tests
  • Added coverage verifying that custom memory resources are used during mark joins.

Walkthrough

mark_join constructors now accept an optional device memory resource, propagate it through the public and internal implementations, and use it for bucket and bloom-filter allocations. A test verifies allocations through a statistics resource adaptor.

Changes

Mark join memory resource support

Layer / File(s)	Summary
Constructor resource plumbing cpp/include/cudf/join/mark_join.hpp, cpp/src/join/mark_join.cuh, cpp/src/join/mark_join.cu	Public and internal constructors add the device-accessible memory resource parameter, default it to the current device resource, and forward it to the implementation.
Resource-backed allocation validation cpp/src/join/mark_join.cu, cpp/tests/join/semi_anti_join_tests.cpp	Bucket and bloom-filter allocators use the supplied resource, and a test confirms nonzero peak allocation through a statistics adaptor.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• rapidsai/cudf#23223: Extends a related hash join constructor with a device memory resource and threads it into internal allocation.

Suggested labels: improvement

Suggested reviewers: davidwendt, wence-, mhaseeb123

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: adding a memory resource parameter to mark_join constructors.
Description check	✅ Passed	The description directly explains the MR addition and its purpose, and matches the code changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

🧹 Nitpick comments (1)

cpp/tests/join/semi_anti_join_tests.cpp (1)

519-531: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Cover the load-factor constructor as well.

This test exercises only the first public constructor; the load-factor overload has an independent mr forwarding path. Add a second case or parameterize the test to cover both.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@cpp/tests/join/semi_anti_join_tests.cpp` around lines 519 - 531, Extend
MarkJoinMemoryResource to also construct cudf::mark_join through the public
load-factor overload, passing the same memory resource and valid load-factor
value, then verify the resource records allocation for that path as well.
Preserve the existing constructor coverage and assertions.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@cpp/tests/join/semi_anti_join_tests.cpp`:
- Around line 519-531: Extend MarkJoinMemoryResource to also construct
cudf::mark_join through the public load-factor overload, passing the same memory
resource and valid load-factor value, then verify the resource records
allocation for that path as well. Preserve the existing constructor coverage and
assertions.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 487ee9a9-6357-424d-8762-0d3310d6b6fd

📥 Commits

Reviewing files that changed from the base of the PR and between 783e214 and 0ae3e47.

📒 Files selected for processing (4)

• cpp/include/cudf/join/mark_join.hpp
• cpp/src/join/mark_join.cu
• cpp/src/join/mark_join.cuh
• cpp/tests/join/semi_anti_join_tests.cpp

[vyasr]

/merge