Make sure docs readers aren't confused by assert_writeable_eq

[Manishearth]

This came up in #5896, around to_parts specifically.

A lot of our doctests use assert_writeable_eq. This is great; it's relatively compact, and in as much as showing the user "the output of this function is this string" it does wonderfully.

However; almost no application code is going to be asserting the result of formatting. Most applications want a string or something. So someone seeing these docs may come away not actually knowing how to obtain what they want. Most of these types have a to_string() so it's not a huge deal when it comes to basic writing, but we could probably have more signposting especially around write_to_parts.

Various suggestions:

• Make sure all writeable types have a to_string. I think we already do this.
• Make sure the docs of the to_string mention Writeable, and mention that there's further functionality on that trait.
• Code that supports parts-formatting should explicitly document it, and have a line or two linking to PartsWrite and such
• assert_writeable_parts_eq should mention PartsWrite and write_to_parts explicitly.

Not any kind of blocker, just docs improvements that we should make at some point.

cc @sffc