Add img to replaced tags which get preserved in HTML from slicing.

[andreip]

fixes #22

Couldn't think of a different approach, since an img isn't really a block, so it'll never have a text within it, so no point in generating a different html in get_line_info functions. Instead, what was missing was it being treated as a special case: don't want to slice a line from the HTML by just looking at the plain text lines, since that could slice an img, need to also look at the start/end refs for replaced tags.

See more about a replaced element (https://developer.mozilla.org/en-US/docs/Web/CSS/Replaced_element). I think it might be worth adding a few more things to the list? e.g. video, embed etc. ; not sure about iframe and how that'd be treated in lxml parsing though, but I suppose you could have an iframe with just an image in it, in which case you'd still want to keep it?

Full list would be a total of 9 replaced elements (or 10 if we also count input; although I'm not sure of all examples where that'd generate sth even if it apparently has no text in it).

[afzalIbnSH]

Can this please be merged?

[afzalIbnSH]

I tested this PR in my project and it works nicely. Would be great if this can be merged. @wojcikstefan

[thomasst]

This looks like a good start, but is still incomplete and not catching all of the cases:

For example, the following test string

html = "Test 2.<br><img src=\"top-image\"><br>On Jun 05, 2018, at 09:56 AM, John Doe &lt;[email protected]&gt; wrote:<br><blockquote><img src=\"https://example.com\" class=\"fr-fic fr-dib\"><br>Some text 1.<br><br>Bart</blockquote><img src=\"bottom-image\">"

should be parsed into:

  {'date': 'Jun 05, 2018, at 09:56 AM',
   'from': 'John Doe <[email protected]>',
   'html': '<div><img src="https://example.com" class="fr-fic fr-dib"><br>Some '
           'text 1.<br><br>Bart</div>',
   'html_bottom': '<img src="bottom-image">',
   'html_top': 'Test 2.<br><img src="top-image">',
   'type': 'reply'}

In this PR, it ignores the top-image (rather than having it in html_top) and includes the bottom-image as part of the html (rather than html_bottom).