Fix indent v2

[cmrschwarz]

This is my second attempt at fixing the more involved indenting issues first brought up in #650 .

It turns out that the Indent node is simply not powerful enough to handle all cases :(.

We will unfortunately have to dynamically track whether a partial ended on a newline or not.
This PR does exactly that.

Here are some of the more difficult edge cases that this solves:

Input

{{#*inline "empty_partial"}}{{/inline}}
<div>
    {{> empty_partial}}
</div>

Output

<div>
</div>

---

Input

{{#*inline "partial"}}foo{{/inline}}
<div>
    {{> partial}}
</div>

Output

<div>
    foo</div>

(yes, this is indeed correct according to handlebars js)

---

Input

Json Input: {"a": "line\ndynamic_trailing_newline\n"}

{{#*inline "dynamic_partial"}}{{a}}{{/inline}}
<div>
    {{> dynamic_partial}}
</div>

Output

<div>
    line
    dynamic_trailing_newline
</div>

[coveralls]

coverage: 81.605% (+0.1%) from 81.461%
when pulling bc67e05 on cmrschwarz:fix_indent_v2
into 3504412 on sunng87:master.

[coveralls]

coverage: 81.625% (+0.2%) from 81.461%
when pulling 16109f6 on cmrschwarz:fix_indent_v2
into 3504412 on sunng87:master.

[sunng87]

I have been super busy these days. Really need some time to deep dive into this. But the test cases look good. I think this approach is more promising.

[sunng87]

I think we can make this field private, as well as the one for DecoratorTemplate

[cmrschwarz]

Sure! I need them in render_helper, so I guess pub(crate) it is.

[coveralls]

coverage: 81.625% (+0.2%) from 81.461%
when pulling 91579a4 on cmrschwarz:fix_indent_v2
into 3504412 on sunng87:master.

[cmrschwarz]

I've been using this in practice for a bit more now and found two minor issues with it.
The commits just now fixed those and added regression tests.
Otherwise this has worked nicely, indenting seems just about perfect now.

Edit: Well ok, you always miss something I guess :). Similar fix for partials below.

[coveralls]

coverage: 81.714% (+0.3%) from 81.461%
when pulling a165d99 on cmrschwarz:fix_indent_v2
into 3504412 on sunng87:master.

[coveralls]

coverage: 81.724% (+0.3%) from 81.461%
when pulling 0af5ab5 on cmrschwarz:fix_indent_v2
into 3504412 on sunng87:master.

[sunng87]

This looks good to me. Thank you for the fix! I have been struggling with those indent for quite a while. This approach seems practical and also keeps complexity at a acceptable level.

[sunng87]

I just tried the case reported in #611, and it seems we still this corner case to cover.

[cmrschwarz]

I just tried the case reported in #611, and it seems we still this corner case to cover.

I'm happy to look into that, though I'm quite busy this week so it might have to wait until the weekend.

[sunng87]

Thank you and no rush @cmrschwarz . Just take your time.