Clarify the the role of utf-8

In several places nbformat seems to choose utf-8 as the default encoding (in particular when `read` or `write` get filenames as input).

Does that mean that notebook files must be utf-8 encoded? If yes, perhaps it would be worth to state it explicitly.

See https://github.com/jupyter/jupyter-sphinx/pull/125 for an example context where this matters.

The issue names no files or tests; start by tracing the filename-based `read` and `write` entry points and reviewing their documentation. Confirm the documented UTF-8 behavior for notebook files, then update the relevant documentation so the encoding requirement or default is explicit.

Written by the indexing model from the issue text.