fix(#1046): support eo-parser 0.61.3

[theaflit]

Closes #1046.

Changes

• bump eo-parser from 0.61.1 to 0.61.3;
• adapt wrong-test-order to the reordered XMIR structure;
• determine source order using the line attributes instead of XML sibling order.

Verification

• mvn "-Dtest=LtByXslTest#testsAllLintsByEo" -DskipITs test
• 339 tests passed, 0 failures, 0 errors.

[theaflit]

@volodya-lombrozo xcop is still failing
The files are not modified in this PR, and the same failure is reproducible on the current master branch. The issue is tracked separately in #1053.
The changes in this PR are limited to the eo-parser 0.61.3 update and wrong-test-order fixtures.

[theaflit]

@volodya-lombrozo All checks are green and the previous review comment has been addressed. Could you please review the latest version when you have a chance?

[yegor256]

@volodya-lombrozo ping

[volodya-lombrozo]

@theaflit Can we render elements in the input to make this test pass?

[theaflit]

@volodya-lombrozo I tried rendering every element in the input by giving bar,
+runs-program, and boom a body.

However, with eo-parser 0.61.3, the resulting XMIR sibling order is
still changed from:

bar, +runs-program, boom

to:

bar, boom, +runs-program

Therefore, this approach still produces no defect. Unless you had a
different way of rendering the elements in mind, it seems that
document is needed here to preserve the sibling order tested by
this fixture.

[volodya-lombrozo]

@theaflit There is a fresh version of EO parser, maybe it's better to use it now?