track fcntl locks

[roblatham00]

tracking counts and kinds of locks issued and time spent obtaining the locks

[wkliao]

CI failed messages: (also reproducible when running command make check locally)

/darshan-accumulator/inject-shared-file-records
...
Error: ../../darshan/darshan-util/tests/unit-tests/darshan-accumulator.c:255: as
sertion failed: 5 == 4 (5 == 4)
...
Error: ../../darshan/darshan-util/tests/unit-tests/darshan-accumulator.c:531: as
sertion failed: 5 == 4 (5 == 4)

Changing == to >= in those 2 lines to the followings seems to resolve this error.

  -    munit_assert_int(DARSHAN_POSIX_VER, ==, 4);
  +    munit_assert_int(DARSHAN_POSIX_VER, >=, 4);

[roblatham00]

I did bump the DARSHAN_POSIX_VER to add the new FCNTL fields.... but I don't yet understand how the test suite is combining records. Well, since I merely added new records and did not modify existing ones I have left the body of the tests the same

[wkliao]

The Python Testing error looks like the same problem I encountered in PR #1102, which adds
a new field, thread ID. My understanding of test_pydarshn is it reads a few darshan log files
in an older version and checks the fields (as DataFrame columns). The expected fields are
hard-coded in the python tests programs. The test will error out if they do not exactly match.

In #1102, you can check my changes to file darshan-util/pydarshan/darshan/backend/cffi_backend.py.
I added checks of the version numbers and adjust the columns accordingly.