[work-in-progress]: Perform all decoding internally

[dralley]

• The data being streamed from the reader would be converted from whichever encoding it is currently in to UTF-8 when copied to a buffer, or else validated as UTF-8 if that is the appropriate encoding, using streaming APIs
• UTF-8 would be the only encoding used within the library itself
• All decoding functions would be removed from the API
• quick-xml would never have to allocate during parsing except from resizing a buffer or doing (un)escaping
• All API functions would be changed to return &str / String / Cow<str> rather than raw bytes
• Known character boundaries can be used to ensure the safety of unsafe fn std::str::from_utf8_unchecked(), which would not incur any overhead

[dralley]

@Mingun seriously I need some feedback on the plan. It would conflict with or render pointless a lot of other changes.

[Mingun]

I'm going to sleep now, I'll give my thoughts later. General note: I prefer moving by small steps, first preparing the necessary infrastructure. To begin with, I suggest looking at #421 first

[codecov-commenter]

Codecov Report

Merging #420 (0bb9922) into master (d34b779) will decrease coverage by 1.37%.
The diff coverage is 74.03%.

@@            Coverage Diff             @@
##           master     #420      +/-   ##
==========================================
- Coverage   49.53%   48.16%   -1.38%     
==========================================
  Files          22       22              
  Lines       13856    13673     -183     
==========================================
- Hits         6864     6586     -278     
- Misses       6992     7087      +95     

Flag	Coverage Δ
unittests	48.16% <74.03%> (-1.38%)	⬇️

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

Impacted Files	Coverage Δ
benches/macrobenches.rs	0.00% <0.00%> (ø)
benches/microbenches.rs	0.00% <0.00%> (ø)
examples/custom_entities.rs	0.00% <0.00%> (ø)
src/lib.rs	9.14% <0.00%> (-3.19%)	⬇️
src/de/escape.rs	19.84% <33.33%> (-1.21%)	⬇️
src/de/var.rs	81.63% <33.33%> (-3.48%)	⬇️
src/reader.rs	57.69% <65.00%> (-2.44%)	⬇️
src/de/mod.rs	77.00% <71.87%> (-1.33%)	⬇️
src/de/map.rs	73.01% <75.00%> (+0.45%)	⬆️
src/events/mod.rs	70.15% <76.47%> (+1.48%)	⬆️
... and 14 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d34b779...0bb9922. Read the comment docs.

[dralley]

I would be fine with merging the commits incrementally.

[dralley]

@Mingun These 5 commits are "ready", if you would like to review them now. There is a lot of temporary hacks in here that are already gone in the full patch series, but it would be an extra 1000+ lines added to this, and big PRs suck to review.

There are more patches here which mostly complete the transition to utf8 types, however, BytesTextStart is a little problematic and I'm not sure what to do with it yet. If you have any ideas, let me know.

[dralley]

Most of these temporaries are already gone later in the patch series, but as I said, it's another 1000 lines of changes

[dralley]

Changing approach, working on top of the first half of the async PR.