bcc/BPF_TABLE can be atomically cleanup?

When i processed `./funclatency.py '/home/zcy/test/a.out:test_func' -Tui 1`. I found that the number of times the function was counted was not the same as the number of times it was executed

```
for (i = 0; i < 1000000; i++)
test_func();
```

```
./funclatency.py '/home/zcy/test/a.out:test_func' -Tui 1
Tracing 1 functions for "/home/zcy/test/a.out:test_func"... Hit Ctrl-C to end.

19:00:55

19:00:56

Function = test_func [60802]
usecs : count distribution
0 -> 1 : 102733 |****************************************|
2 -> 3 : 134 | |
4 -> 7 : 346 | |
8 -> 15 : 15 | |
16 -> 31 : 77 | |
32 -> 63 : 64 | |
64 -> 127 : 44 | |
128 -> 255 : 14 | |
256 -> 511 : 1 | |

avg = 1 usecs, total: 146898 usecs, count: 103428

19:00:58

Function = test_func [60802]
usecs : count distribution
0 -> 1 : 398470 |****************************************|
2 -> 3 : 554 | |
4 -> 7 : 1588 | |
8 -> 15 : 108 | |
16 -> 31 : 301 | |
32 -> 63 : 219 | |
64 -> 127 : 151 | |
128 -> 255 : 51 | |
256 -> 511 : 6 | |

avg = 1 usecs, total: 561911 usecs, count: 396896

19:00:59

Function = test_func [60802]
usecs : count distribution
0 -> 1 : 379875 |****************************************|
2 -> 3 : 579 | |
4 -> 7 : 1465 | |
8 -> 15 : 67 | |
16 -> 31 : 309 | |
32 -> 63 : 226 | |
64 -> 127 : 145 | |
128 -> 255 : 60 | |
256 -> 511 : 5 | |

avg = 1 usecs, total: 539264 usecs, count: 378027

19:01:00

19:01:01
```

```
103428+396896+378027 != 1000000
```

Start by reproducing the reported command with funclatency.py and the provided million-call test. Inspect how BPF_TABLE data is cleaned up and counted between output intervals; done means determining why the reported counts differ from executions and establishing atomic cleanup or documenting the limitation.

Written by the indexing model from the issue text.