Fix of instructions.rs comment, to_precision impl and rfc changes

[NathanRoyer]

This Pull Request fixes/closes #1038.

It changes the following:

• applying the changes of RFC 2700
• fixing the comment mistake from Seemingly wrong comment for Rational in instruction enum #1038
• fixing my previously incorrect implementation of Number.prototype.toPrecision

[RageKnify]

Code looks good, but I think it'd be nice to have 1 or 2 test cases for the change you did to round_to_precision to make sure no regressions are introduced in the future. Is it easy to do that @NathanRoyer?

[RageKnify]

test262 numbers actually look bad

Test262 conformance changes:

Test result	master count	PR count	difference
Total	78,497	78,497	0
Passed	24,986	24,984	-2
Ignored	15,587	15,587	0
Failed	37,924	37,926	+2
Panics	16	16	0
Conformance	31.83	31.83	-0.00%

[Razican]

This would need a rebase in order for the build to pass.

[RageKnify]

@NathanRoyer this directory has the test262 tests for Number.prototype.toPrecision, can you take a look to try to understand what is now failing that wasn't before? You could also use the test runner if you prefer.

[Razican]

I suggest changing a couple of things. The code looks good, but as @RageKnify mentions, you should see why there are 2 new tests failing.

Also, it needs a rebase, again.

[0x7D2B]

Latest test262 results:

Test262 conformance changes:

Test result	master count	PR count	difference
Total	78,873	78,873	0
Passed	26,498	26,502	+4
Ignored	15,604	15,604	0
Failed	36,771	36,767	-4
Panics	27	27	0
Conformance	33.60%	33.60%	+0.01%

It doesn't post the comment but you can find the text in the Update Comment section of test262 results: https://github.com/boa-dev/boa/pull/1135/checks?check_run_id=2649202840

[Razican]

Benchmarks look pretty good, and conformance too. It's good to merge from my side :) thank you!