Do not write excess indents arount text fields when serialize with serde

[Mingun]

This PR alternative and more full implementation of the fix for excess indents around text fields. It implements the Python scheme of writing indents:

struct Xml {
  before: String,// = "value1"
  #[serde(rename = "$text")] // or $value
  text: String,// = "text"
  after: String,// = "value2"
}

before, $text and after

<Xml>
  <before>value1</before>text<after>value2</after>
</Xml>

$text and after

<Xml>text<after>value2</after>
</Xml>

before and $text

<Xml>
  <before>value1</before>text</Xml>

$text only

<Xml>text</Xml>

The implementation never write indent after Option fields, because Option can contain a type that is sensible to the surrounding text content, for example Option<String>. Because serde does not provide the type of None when serialize it, it is impossible to be smart here and write indent in one cases but not other, so I choosed the safe variant.

As a side effect, to_writer, to_writer_with_root and Serializer not returns bool on success. The true means that is will be safe to write text (in particular, indent) after the serialized object, it will not be interpreted as a part of content during deserialization. If false will be returned, then writing additional text content may change data what deserializer will see.

Closes #655, closes #814

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

Attention: Patch coverage is 98.75000% with 1 line in your changes missing coverage. Please review.

Project coverage is 60.19%. Comparing base (39b5905) to head (da88af2).
Report is 14 commits behind head on master.

Files with missing lines	Patch %	Lines
src/se/content.rs	97.43%	1 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #820      +/-   ##
==========================================
+ Coverage   60.08%   60.19%   +0.10%     
==========================================
  Files          41       41              
  Lines       15975    15999      +24     
==========================================
+ Hits         9599     9630      +31     
+ Misses       6376     6369       -7     

Flag	Coverage Δ
unittests	60.19% <98.75%> (+0.10%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Mingun]

Updated: even better solution. Now serialization returns the classification result (instead of just boolean) and you can decide what to do with it.

[dralley]

How is vec![1, 2, 3] distinguished from vec![123]

[Mingun]

Unfortunately, in any way. This is what happens now, if you do not use indents.

[Mingun]

In #823 I changed the return value in that case to error

[dralley]

I'm definitely missing something here.

How would Tuple("first42", 0) be distinguished from Tuple("first", 420)

[Mingun]

The same answer as above. This is restriction of current implementation. I have a TODO to forbid serialization consequent text fields:

quick-xml/src/se/content.rs

Line 75 in 6eea6bb

//TODO: add settings to disallow consequent serialization of primitives

[dralley]

Is there an issue written up for that already, and if not can you write one and reference it

[Mingun]

No, there is no issue yet

[Mingun]

In #823 I changed the return value in that case to error