Use a more efficient xml parser library

[nineinchnick]

What

Reduce the memory usage and allow processing large XML reports (>100 MB) on public GHA runners, that have only 7 GB of memory.

How

Replace xml-js with libxmljs.

I tested memory usage using:

gtime -f '\t%E real,\t%U user,\t%S sys,\t%K amem,\t%M mmem' npm run test

Before:

	0:18.49 real,	16.05 user,	10.00 sys,	0 amem,	230432 mmem

After:

	0:19.70 real,	16.59 user,	10.51 sys,	0 amem,	225952 mmem

Here's another test with a large XML file. This is a test report from the Trino project, it's available in the CI as artifacts.

export INPUT_SKIP_PUBLISHING=true
export INPUT_REPORT_PATHS=$HOME/Downloads/test-reports/test\ report\ test-other-modules/lib/trino-parquet/target/surefire-reports/TEST-TestSuite.xml

The file is almost 100 MB:

% du -mc "$INPUT_REPORT_PATHS" 
89	/Users/jwas/Downloads/test-reports/test report test-other-modules/lib/trino-parquet/target/surefire-reports/TEST-TestSuite.xml

Now run the action:

% gtime -f '\t%E real,\t%U user,\t%S sys,\t%K amem,\t%M mmem' node index.js

Before:

	0:08.11 real,	12.47 user,	1.83 sys,	0 amem,	3198624 mmem

After:

	0:03.00 real,	2.69 user,	0.11 sys,	0 amem,	694784 mmem

3.2 GB drops to 0.7 GB. There are also savings in processing time.

The changes in tests were required because:

• in some tests for utils.js, the stacktrace was not properly set
• tabs in attributes should be encoded, otherwise they get converted to spaces, which conforms to XML standards

[ghaiszaher]

@nineinchnick could you fix the lint problems npm run estlint?

[nineinchnick]

@ghaiszaher done! I bumped ecmaVersion from 2018 to 2020 to support the optional chaining operator. I think Node 16 should support most of 2020: https://node.green/#ES2020

[ghaiszaher]

Thanks for your contribution!