Remove mr from test fixtures

[vyasr]

Description

This is a follow-up to #14057. The PR was split in two because in that PR this particular change led to an unexpected and very specific test failure (see #14057 (comment)). If this PR reproduces that issue we will need to debug more closely on the appropriate hardware.

Checklist

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

[karthikeyann]

Change looks good to me!

[karthikeyann]

Same tests failed as https://github.com/rapidsai/cudf/actions/runs/6126279847/job/16630671974?pr=14057
doesn't seem random, but can't find a pattern either.
No errors while running it with valgrind, compute-sanitizer (memcheck, racecheck, initcheck, synccheck)

[  FAILED  ] 5 tests, listed below:
[  FAILED  ] RoundRobinTest/4.RoundRobinNPartitionsDivideNRows, where TypeParam = unsigned char
[  FAILED  ] RoundRobinTest/7.RoundRobinNPartitionsDivideNRows, where TypeParam = unsigned long
[  FAILED  ] RoundRobinTest/10.RoundRobinNPartitionsDivideNRows, where TypeParam = double
[  FAILED  ] RoundRobinTest/15.RoundRobinNPartitionsDivideNRows, where TypeParam = cuda::std::__4::chrono::time_point<cuda::std::__4::chrono::system_clock,cuda::std::__4::chrono::duration<long,cuda::std::__4::ratio<1l,1000000000l> > >
[  FAILED  ] RoundRobinTest/23.RoundRobinNPartitionsDivideNRows, where TypeParam = numeric::fixed_point<__int128,(numeric::Radix)10>

 5 FAILED TESTS

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[karthikeyann]

/ok to test

[vyasr]

@robertmaynard and I have both spent some time trying to investigate the issues here. We were unable to reproduce on at least one other arm node, and Robert said he was not able to reproduce this using CUDA 12. Since the issue appears to be very specific and compiler-related, I'm going to close this PR for now. The plan will be to reopen this once we drop CUDA 11 support since the issue appears to be specific to CUDA 11.

[vyasr]

Tracking in #14236