Frequent (but nondeterministic) failures in test_profiling in Frontier CI

[elliottslaughter]

As of master commit a97cfa564afe8050ae5c455109bc4ac774916e66, I have started seeing frequent, but nondeterministic, failures in test_profiling on Frontier CI. (Note: I don't believe this is related to that specific commit, it's just the one where I happen to see the failures begin to occur.)

Sample failures: 1, 2, 3, 4, 5

Sample success: 1

Failure output looks like:

 58/109 Test  #58: test_profiling ...................***Failed   13.73 sec
top level task - getting machine and list of CPUs
[0 - 7fffe588e340]    0.159977 {3}{app}: profiling response task on processor 1d00000000000001
got profiling response - 136 bytes
Bytes: 04 00 00 00 00 00 00 00 03 00 00 00 04 00 00 00 13 00 00 00 28 00 00 00 38 00 00 00 60 00 00 00 68 00 00 00 78 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 99 49 8c 03 00 00 00 00 33 4d 8c 03 00 00 00 00 5e a2 8d 03 00 00 00 00 6d 7c 85 09 00 00 00 00 f6 cd 87 09 00 00 00 00 00 00 00 00 00 00 00 00 1c c2 8d 03 00 00 00 00 00 00 00 00 00 00 00 80 01 00 00 00 00 00 00 00 02 00 00 00 00 03 00 1d
op status = 0 (code = 0, details = 0 bytes)
op timeline = 59526451 59613790 159743085 159895030 (87339 100129295 151945)
op gpu timeline = 59621916 -9223372036854775808 (9223372036795153892)
test_profiling: /lustre/orion/ums036/proj-shared/ci/38148_79942/test/realm/test_profiling.cc:196: void response_task(const void*, size_t, const void*, size_t, Realm::Processor): Assertion `(result == OperationStatus::TERMINATED_EARLY) || (op_timeline->start_time <= op_timeline->end_time)' failed.
Signal 6 received by node 0, process 2489157 (thread 7fffe588e340) - obtaining backtrace
Signal 6 received by process 2489157 (thread 7fffe588e340) at: stack trace: 11 frames
  [0] = /lib64/libc.so.6(gsignal+0x10d) [0x7ffff3c5ed2b]
  [1] = /lib64/libc.so.6(abort+0x176) [0x7ffff3c603e4]
  [2] = /lib64/libc.so.6(+0x42c69) [0x7ffff3c56c69]
  [3] = /lib64/libc.so.6(__assert_fail+0x43) [0x7ffff3c56cf1]
  [4] = /lustre/orion/ums036/proj-shared/ci/38148_79942/tmpuoldqey0/build/bin/test_profiling(response_task(void const*, unsigned long, void const*, unsigned long, Realm::Processor)+0xe8e) [0x48aaae]
  [5] = /lustre/orion/ums036/proj-shared/ci/38148_79942/tmpuoldqey0/build/bin/test_profiling() [0x68dc18]
  [6] = /lustre/orion/ums036/proj-shared/ci/38148_79942/tmpuoldqey0/build/bin/test_profiling() [0x68dcb5]
  [7] = /lustre/orion/ums036/proj-shared/ci/38148_79942/tmpuoldqey0/build/bin/test_profiling() [0x68c12e]
  [8] = /lustre/orion/ums036/proj-shared/ci/38148_79942/tmpuoldqey0/build/bin/test_profiling() [0x6930e9]
  [9] = /lib64/libc.so.6(+0x6179d) [0x7ffff3c7579d]
  [10] = [(nil)]

This is now breaking all of our Frontier CI builds, so it would be nice if we could get this fixed.

@eddy16112 @apryakhin