`test_blockwise_dataframe_io[True-False-hdf]` is flaky on osx CI

I noticed this test periodically failing due to a pytables locking issue today (may have started earlier).

Traceback

```
_________________ test_blockwise_dataframe_io[True-False-hdf] __________________
[gw0] darwin -- Python 3.8.12 /Users/runner/miniconda3/envs/test-environment/bin/python

c =
tmpdir = local('/private/var/folders/24/8k48jl6d249_n_qfxwsl6xvm0000gn/T/pytest-of-runner/pytest-0/popen-gw0/test_blockwise_dataframe_io_Tr3')
io = 'hdf', fuse = False, from_futures = True

@pytest.mark.filterwarnings(
"ignore:Running on a single-machine scheduler when a distributed client "
"is active might lead to unexpected results."
)
@pytest.mark.parametrize(
"io",
["parquet-pyarrow", "parquet-fastparquet", "csv", "hdf"],
)
@pytest.mark.parametrize("fuse", [True, False, None])
@pytest.mark.parametrize("from_futures", [True, False])
def test_blockwise_dataframe_io(c, tmpdir, io, fuse, from_futures):
pd = pytest.importorskip("pandas")
dd = pytest.importorskip("dask.dataframe")

df = pd.DataFrame({"x": [1, 2, 3] * 5, "y": range(15)})

if from_futures:
parts = [df.iloc[:5], df.iloc[5:10], df.iloc[10:15]]
futs = c.scatter(parts)
ddf0 = dd.from_delayed(futs, meta=parts[0])
else:
ddf0 = dd.from_pandas(df, npartitions=3)

if io.startswith("parquet"):
if io == "parquet-pyarrow":
pytest.importorskip("pyarrow.parquet")
engine = "pyarrow"
else:
pytest.importorskip("fastparquet")
engine = "fastparquet"
ddf0.to_parquet(str(tmpdir), engine=engine)
ddf = dd.read_parquet(str(tmpdir), engine=engine)
elif io == "csv":
ddf0.to_csv(str(tmpdir), index=False)
ddf = dd.read_csv(os.path.join(str(tmpdir), "*"))
elif io == "hdf":
pytest.importorskip("tables")
fn = str(tmpdir.join("h5"))
> ddf0.to_hdf(fn, "/data*")

dask/tests/test_distributed.py:370:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
dask/dataframe/core.py:1620: in to_hdf
return to_hdf(self, path_or_buf, key, mode, append, **kwargs)
dask/dataframe/io/hdf.py:251: in to_hdf
compute_as_if_collection(
dask/base.py:319: in compute_as_if_collection
return schedule(dsk2, keys, **kwargs)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/client.py:3015: in get
results = self.gather(packed, asynchronous=asynchronous, direct=direct)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/client.py:2167: in gather
return self.sync(
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/utils.py:311: in sync
return sync(
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/utils.py:378: in sync
raise exc.with_traceback(tb)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/utils.py:351: in f
result = yield future
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/tornado/gen.py:762: in run
value = future.result()
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/distributed/client.py:2030: in _gather
raise exception.with_traceback(traceback)
dask/dataframe/io/hdf.py:27: in _pd_to_hdf
pd_to_hdf(*args, **kwargs)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/pandas/core/generic.py:2606: in to_hdf
pytables.to_hdf(
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/pandas/io/pytables.py:277: in to_hdf
with HDFStore(
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/pandas/io/pytables.py:561: in __init__
self.open(mode=mode, **kwargs)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/pandas/io/pytables.py:710: in open
self._handle = tables.open_file(self._path, self._mode, **kwargs)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/tables/file.py:300: in open_file
return File(filename, mode, title, root_uep, filters, **kwargs)
../../../miniconda3/envs/test-environment/lib/python3.8/site-packages/tables/file.py:750: in __init__
self._g_new(filename, mode, **params)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

> ???
E tables.exceptions.HDF5ExtError: HDF5 error back trace
E
E File "H5F.c", line 620, in H5Fopen
E unable to open file
E File "H5VLcallback.c", line 3502, in H5VL_file_open
E failed to iterate over available VOL connector plugins
E File "H5PLpath.c", line 579, in H5PL__path_table_iterate
E can't iterate over plugins in plugin path '(null)'
E File "H5PLpath.c", line 620, in H5PL__path_table_iterate_process_path
E can't open directory: /Users/runner/miniconda3/envs/test-environment/lib/hdf5/plugin
E File "H5VLcallback.c", line 3351, in H5VL__file_open
E open failed
E File "H5VLnative_file.c", line 97, in H5VL__native_file_open
E unable to open file
E File "H5Fint.c", line 1898, in H5F_open
E unable to lock the file
E File "H5FD.c", line 1625, in H5FD_lock
E driver lock request failed
E File "H5FDsec2.c", line 1002, in H5FD__sec2_lock
E unable to lock file, errno = 35, error message = 'Resource temporarily unavailable'
E
E End of HDF5 error back trace
E
E Unable to open/create file '/private/var/folders/24/8k48jl6d249_n_qfxwsl6xvm0000gn/T/pytest-of-runner/pytest-0/popen-gw0/test_blockwise_dataframe_io_Tr3/h5'

tables/hdf5extension.pyx:486: HDF5ExtError
```

Log from failing test: https://github.com/dask/dask/runs/5558289646?check_suite_focus=true#step:6:21912

Start with dask/tests/test_distributed.py and run the test_blockwise_dataframe_io[True-False-hdf] case on macOS, using the linked CI log as the failure reference. Read dask/dataframe/io/hdf.py around _pd_to_hdf and trace the PyTables locking failure. Done means the HDF case no longer flakes on macOS CI and the relevant test still passes.

Written by the indexing model from the issue text.