Intermittent `KeyError` when writing to S3

Still experiencing intermittent Access Denied errors see #218.

However today a new one has popped up when writing to_parquet. In this case:

df.to_parquet(
    'path/to/20190719.parquet',
    engine='fastparquet',
    compression='snappy',
    times='int96',  # spark supported timestamps  
    storage_options={'s3_additional_kwargs': server_side_encryption_args},
)

occasionally (with a similar freq to access denied so far) is giving:

'path/to/20190719.parquet'
Traceback (most recent call last):
  File "<ipython-input-50-e89c9bf2fe97>", line 16, in <module>
    storage_options={'s3_additional_kwargs': sse_args},
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/dask/dataframe/core.py", line 3618, in to_parquet
    return to_parquet(self, path, *args, **kwargs)
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/dask/dataframe/io/parquet.py", line 1496, in to_parquet
    out.compute()
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/dask/base.py", line 175, in compute
    (result,) = compute(self, traverse=False, **kwargs)
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/dask/base.py", line 446, in compute
    results = schedule(dsk, keys, **kwargs)
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/client.py", line 2527, in get
    results = self.gather(packed, asynchronous=asynchronous, direct=direct)
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/client.py", line 1823, in gather
    asynchronous=asynchronous,
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/client.py", line 763, in sync
    self.loop, func, *args, callback_timeout=callback_timeout, **kwargs
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/utils.py", line 332, in sync
    six.reraise(*error[0])
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/six.py", line 693, in reraise
    raise value
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/utils.py", line 317, in f
    result[0] = yield future
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/tornado/gen.py", line 735, in run
    value = future.result()
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/tornado/gen.py", line 742, in run
    yielded = self.gen.throw(*exc_info)  # type: ignore
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/distributed/client.py", line 1680, in _gather
    six.reraise(type(exception), exception, traceback)
  File "/mnt/miniconda/envs/jestr-etl/lib/python3.7/site-packages/six.py", line 692, in reraise
    raise value.with_traceback(tb)
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/dask/dataframe/io/parquet.py", line 617, in _read_parquet_row_group
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/fastparquet/core.py", line 302, in read_row_group_file
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/fsspec/spec.py", line 689, in open
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 303, in _open
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 920, in __init__
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/fsspec/spec.py", line 884, in __init__
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 479, in info
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/fsspec/spec.py", line 495, in info
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 497, in ls
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 430, in _ls
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1566335097751_3990/container_1566335097751_3990_01_000015/environment/lib/python3.7/site-packages/s3fs/core.py", line 339, in _lsdir
KeyError: 'path/to/20190719.parquet'

environment is:

• dask 2.1.0
• distributed 2.1.0
• s3fs 3.3.0
• fsspec 0.4.3

Reproduce the intermittent failure with the listed dask, distributed, s3fs, and fsspec versions using the shown to_parquet call. Start in s3fs/core.py at _lsdir and follow the fsspec info/open path from the traceback; done means the KeyError no longer occurs for this S3 write/read flow and the behavior is covered by a regression test.