hdf5 formatter cannot write a dataset twice

Nobody has claimed this yet.

When calling the `write_copy` method of the dataset on an existing location an error is generated.

A minimal example:

```
import numpy as np
import qcodes
from qcodes.tests.data_mocks import DataSet2D
from qcodes.data.hdf5_format import HDF5Format

qcodes.DataSet.default_formatter = HDF5Format()

dataset = DataSet2D()
dataset.write()
newlocation = dataset.location
dataset.write_copy(location=newlocation)
```
And another variation
```
from qcodes.tests.data_mocks import DataSet2D
from qcodes.data.hdf5_format import HDF5Format,HDF5FormatMetadata
qcodes.DataSet.default_formatter = HDF5Format()

dataset = DataSet2D()
dataset.write()
dataset.formatter.write(dataset, force_write=True)
```
@core @AdriaanRol

Start by reproducing the two examples with DataSet2D and inspect qcodes/data/hdf5_format.py, especially the formatter write path and write_copy handling for an existing location. Determine why a second write raises an error, then verify that repeated writes to the same dataset location complete successfully without breaking the initial write behavior.

Written by the indexing model from the issue text.