Start cleaning up staging/sm tests

[anba]

Changes:

• Replace harness functions from harness/sm with standard test262 harness functions.
• Remove now unused harness functions from harness/sm.
• Both changes should help a bit for Deal with slow staging/sm tests #4402, because there's now less harness code to execute.
• Avoid testing error messages, because error messages aren't standardised.
• Remove unused files.
• Remove noStrict flag where possible.
• And many more clean-ups and fixes.

I've also identified a couple of tests which should be removed from test262, because they test non-standard behaviour:

• Tests for Function.caller.
• Date.parse with non-standard inputs.
• Tests how accurate Math functions are.
• If the ToString output for bound functions contains the function name.
• Large RegExp tests.
• Cycle detection tests.
• Tests using SpiderMonkey shell functions like disassemble or timeout.

In a follow-up PR I can remove these tests, too.

[ptomato]

Reviewed up to "Remove non262-extensions-shell.js"

[ptomato]

Reviewed up to "Replace makeIterator with inline definitions"

[ptomato]

Reviewed up to "Remove assertThrownErrorContains"

[ptomato]

We discussed this in the test262 maintainers meeting today. We'll just go ahead and merge it; these are staging tests and so should not be held up for this long by review, and the improvements have been good so far. I'll address the existing comments, and there's one more thing I'd like to do — I'd like to convert the BUGNUMBER assignments to comments with links to Mozilla's bug tracker, rather than just removing them, as I think the history can be relevant to future consumers of these tests. After I do that, I'll merge it.

[ptomato]

Stamp

[iamstolis]

The following tests started to fail after this PR:

• staging/sm/Math/acosh-exact.js
• staging/sm/Math/atanh-exact.js
• staging/sm/Math/log1p-exact.js
• staging/sm/Math/trunc.js
  These tests use ONE_MINUS_EPSILON and/or ONE_PLUS_EPSILON but they do not declared dependency on sm/non262-Math-shell.js (where these constants are defined).

I have created #4588 to track this problem.