One call to H5*ARRAYmake() used completely different indentation style.

hdf5 doesn't like that:
h5py/h5py#1948
HDFGroup/hdf5@16349c5

> The 'not suitable for filters' message does seem to have been added
> in 1.12.1 From the code, it looks like it will refuse to apply any
> filters to vlen strings.

This patch is based on the assumption that the hdf5 code is *correct*, and
indeed fletcher32 shouldn't be used in that vlarrays. But if hdf5 is wrong,
then the fix should be their side. It might make sense to apply this to
get the tests passing again, even if ultimately hdf5 is adjusted too.

Fixes PyTables#845.

Add python3.10 classifier

This enables to attach attributes during the assembly of a IsDescription
class. Example:
```
class TableLayout(tables.IsDescription):
    freq = tables.Float32Col(
        attrs={"val": 13.3, "unit": "Hz", "description": "Ref. freq"})
```

…l_attrs

This registers the Col._v_col_attrs in the ColumnAttributeSet and stores them
as table attributes. This way column attributes set in the IsDescription
assembly are propagated to the Column attributes.

Otherwise the attributes are only build when the column is used directly
for attribute assignment. This then does not store the attributes inserted
in the IsDescription class.

Requested updates to merge PyTables PR PyTables#893