Add table_name option to ParsingContext()

[will-moore]

See IDR/idr0079-hartmann-lateralline#5 (comment)

Usage:

    ctx = ParsingContext(
        client, image._obj, file=csv_name, allow_nan=True,
        table_name=table_name
    )
    ctx.parse()

This change used in IDR/idr0079-hartmann-lateralline@57e507f
to create bulk annotations with custom names:

NB: ParsingContext() has several parameters that are not used: fileid, cfg, cfgid, attach, options, batch_size, loops, ms. Seems like these should be removed, although that will break code that still includes them.

[sbesson]

One minor question about the default value. Otherwise this change looks fairly straightforward.

As mentioned in the related PR, I did not find any downstream code that relies on the table name and would be broken by this PR. Instead all downstream code filters by namespace.

Thanks for raising the unused parameters. As it stands, the issue is that omero metadata populate will pass arguments independently of the selected context so there is a need for some redundant declaration at the moment.
In #59 (comment), I suggested that using **kwargs might be an option to be lenient about extra parameters and declare only the relevant parameters across Context implementations.

[sbesson]

Thanks. Only two questions are:

• can we add an integration test to capture the new keyword?
• after that, should we release as 0.9.0 as this is a new backwards compatible API?

[will-moore]

Added tests

[sbesson]

Now released as https://libraries.io/pypi/omero-metadata/0.9.0