Don't include html_top in a no-message forward

[andreip]

Addresses #14

I'm not sure if I fixed the root cause of it, did my best to understand what's happening here.

The top_range in unwrap_html would be an interval like (0, 1) and the trim_slice would make it (1, 1) which I think is wrong to pass it along to slice_tree after.

Since slice_tree now receives a wrong interval since start_refs[1] appears after end_refs[1] so it's quite unusual situation. To be precise, in slice_tree:

>>> et.getelementpath(start_ref[0])
body/div/div/div
>>> et.getelementpath(end_ref[0])
'body/div/div/br[1]'

and by checking the HTML of the test HTMLUnwrapTestCase.test_gmail_forward_no_message you can see what I'm referring to that end appears before the start.

[andreip]

this PR has 2 line changes + the rest are 2 tests, should be pretty easy to follow.

[andreip]

bump

[thomasst]

Thanks for finding this! I believe this should also be handled in in slice_tree (it's similar to https://github.com/andreip/quotequail/blob/7d0cf6a7e2def22a37b65414d3f2d3a7f9cfc410/quotequail/_html.py#L161)

[andreip]

@thomasst you're right, that did fit in slice_tree function better. See updated PR.