Improve pipeline.writer.WriterManager

Nabu uses a wrapper on top on various tomoscan XXVolume : pipeline.writer.WriterManager.

Using tomoscan volumes solves quite a few things in term of API, but there are still multiple issues:

• When using HDF5, we still have to do self.file_prefix += str("_%05d" % self.start_index) because we are writing partial files. This should be done by tomoscan but it's likely incompatible with its current architecture
• _configure_metadata() is too long, and somehow duplicates what is done in LegacyNXProcessWriter.
• LegacyNXProcessWriter still has to be used for writing 1D data (histogram)

Generally speaking, the design is shaky:

• WriterManager tries to use directly XXVolume classes from tomoscan, but there are still many corner case and code to handle it
• In some cases (histogram and .vol), using tomoscan directly is not possible, which entails even more specialized code

A more proper way of doing things would be:

• Define one writer per file type, with the same API, as it was done by nabu.io.writer
• Each writer can be backed by tomoscan volume
• WriterManager does minimal things to assemble them and dispatch parameters