Optimize and reduce memory usage of XML serialization

[nielsdos]

The serialization process uses the system allocator and requires a copy to request allocated memory once finished. This patch improves this by using smart_str to build the resulting string, reducing the number of copies and reducing total peak memory usage.

[devnexen]

How looks like the perf improvement ? :)

[Girgias]

Use size_t instead of int?

[nielsdos]

The interface is defined by libxml, so it has to be int.

[nielsdos]

How looks like the perf improvement ? :)

For 100 serializations:

[niels@ARCH RELx64]$ hyperfine './sapi/cli/php xml.php' './sapi/cli/php_old xml.php'
Benchmark 1: ./sapi/cli/php xml.php
  Time (mean ± σ):     258.7 ms ±   5.9 ms    [User: 250.2 ms, System: 6.6 ms]
  Range (min … max):   252.1 ms … 267.9 ms    11 runs
 
Benchmark 2: ./sapi/cli/php_old xml.php
  Time (mean ± σ):     319.7 ms ±  19.1 ms    [User: 281.9 ms, System: 36.1 ms]
  Range (min … max):   301.9 ms … 368.9 ms    10 runs
 
Summary
  ./sapi/cli/php xml.php ran
    1.24 ± 0.08 times faster than ./sapi/cli/php_old xml.php

As for memory usage:

Before: Maximum resident set size (kbytes): 25880
After: Maximum resident set size (kbytes): 24568

Test case used: https://gist.github.com/nielsdos/369813d1b1c5c146a6fd7992b8ddbc28

[devnexen]

MSTM