feat: add ConvertReader for documents read from an io.Reader

[hownowstephen]

ConvertReader(r io.Reader, opts ...Option) converts a document straight from a reader. Convert now wraps it with a strings.Reader.

text, err := textplain.ConvertReader(part, textplain.WithLinks(textplain.LinksFootnotes))

A mail pipeline usually has the HTML part as a reader, such as a MIME part or a decoded transfer encoding. Callers no longer need to read it into a string first, because html.Parse already consumes a reader.

This also brings coverage to 100%. The html.Parse error return was the last uncovered statement, since a strings.Reader never fails, and now a failing reader reaches it. TestConvertReader uses iotest.ErrReader to check that read errors come back unchanged.

The change is additive, and Convert's signature and behaviour are the same. The README mentions ConvertReader under Usage.