[benchmark] Measure all threads' CPU usage in callback benchmarks (#37936)

google/benchmark only measures the main thread's CPU by default, and benchmark iteration depends on CPU utilization. These benchmarks were taking a long time since the callback API uses a lot of internal threads, and the benchmarks weren't producing useful metrics. See the documentation here https://google.github.io/benchmark/user_guide.html#cpu-timers

Test command for master: `bazel test --config=opt --test_output=streamed //test/cpp/microbenchmarks:bm_callback_streaming_ping_pong --test_arg='--benchmark_min_time=0.05s' --test_arg='--benchmark_min_warmup_time=0.01' --test_arg='--benchmark_filter=.*<MinIn.*/*/1$'`

For this PR, change the filter to `--test_arg='--benchmark_filter=.*<MinIn.*/*/1/proc.*'`

Before this change (scroll right for `inf` bytes/s, and note the 10k iterations even though the specified benchmark time had been reached):
```
---------------------------------------------------------------------------------------------------------------------------------------
Benchmark                                                                             Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------------------------------------------------------------
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/0/1              38932 ns         1052 ns        10000 bytes_per_second=0/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/1/1              39002 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/8/1              39457 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/64/1             51693 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/512/1            41174 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/4096/1           45006 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/32768/1          70990 ns        0.000 ns        10000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/262144/1        302831 ns        0.000 ns         1000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/2097152/1      2202385 ns        0.000 ns         1000 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/16777216/1    30137348 ns        0.000 ns           10 bytes_per_second=inf/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/134217728/1  437821865 ns        0.000 ns            1 bytes_per_second=inf/s
...
//test/cpp/microbenchmarks:bm_callback_streaming_ping_pong               PASSED in 25.7s
```

After this change:

```
--------------------------------------------------------------------------------------------------------------------------------------------------------------
Benchmark                                                                                                    Time             CPU   Iterations UserCounters...
--------------------------------------------------------------------------------------------------------------------------------------------------------------
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/0/1/process_time/real_time              38233 ns        75471 ns         1768 bytes_per_second=0/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/1/1/process_time/real_time              38847 ns        76616 ns         1781 bytes_per_second=50.2778Ki/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/8/1/process_time/real_time              38848 ns        76929 ns         1758 bytes_per_second=402.21Ki/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/64/1/process_time/real_time             49873 ns        95929 ns         1685 bytes_per_second=2.44761Mi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/512/1/process_time/real_time            41606 ns        80703 ns         1431 bytes_per_second=23.4718Mi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/4096/1/process_time/real_time           45495 ns        86097 ns         1335 bytes_per_second=171.722Mi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/32768/1/process_time/real_time          71506 ns       117093 ns          806 bytes_per_second=874.048Mi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/262144/1/process_time/real_time        353012 ns       433389 ns          183 bytes_per_second=1.38319Gi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/2097152/1/process_time/real_time      3059811 ns      3215404 ns           18 bytes_per_second=1.27663Gi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/16777216/1/process_time/real_time    34951210 ns     35279138 ns            2 bytes_per_second=915.562Mi/s
BM_CallbackBidiStreaming<MinInProcess, NoOpMutator, NoOpMutator>/134217728/1/process_time/real_time  443034887 ns    444528037 ns            1 bytes_per_second=577.833Mi/s
...
//test/cpp/microbenchmarks:bm_callback_streaming_ping_pong               PASSED in 5.1s
```

Closes #37936

COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/37936 from drfloob:bm-callback-measures-all-thread-cpu d5ba8420bd
PiperOrigin-RevId: 686640919

test/cpp/microbenchmarks/bm_callback_streaming_ping_pong.cc

@@ -47,6 +47,7 @@ static void StreamingPingPongArgs(benchmark::internal::Benchmark* b) {
        msg_size == 0 ? msg_size++ : msg_size *= 8) {
     b->Args({msg_size, 1});
     b->Args({msg_size, 2});
+    b->MeasureProcessCPUTime()->UseRealTime();
   }
 }