Editorial: clarify string representation of numbers with radix ≠ 10

[michaelficarra]

Fixes #2847 by adding a radix parameter to Number::toString.

/cc @bathos

[bakkot]

LGTM other than comment but I want to wait for @syg's approval as well.

[bathos]

@michaelficarra %Bigint.prototype.toString% uses the same “should be a generalization” text and points at BigInt::toString. As far as I can tell the algorithm-in-practice aligns with Number-when-not-base-10. Would it make sense (and would it be desirable) for them to point to a common algorithm that’s agnostic to the specific numeric language type?

(This PR already stands on its own with plenty of value, so I’m not suggesting it should address anything related to %BigInt.prototype.toString% — just curious about how that could end up looking.)

[gibson042]

I like the improved generality and precision, and this does a decent job of containing the decimal-specific scale weirdness. That said, I do wonder if the latter could be achieved even better, e.g.

1. If _radix_ ≠ 10 or _n_ is in the inclusive interval from -5 to 21, then
  1. If _n_ ≥ _k_, return the string-concatenation of…
  1. If _n_ > 0, return the string-concatenation of…
  1. Else,
    1. Assert: _n_ ≤ 0.
    1. Return the string-concatenation of…
1. Assert: _radix_ is 10.
1. …

[michaelficarra]

@gibson042 Thanks for the suggestions. I took all of them. Please re-review to check my work.

@bathos Agree, we could do something similar about %Bigint.prototype.toString%. Can you open an issue to track it? I don't think they should share an algorithm, but we can at least make a similar clarification.

[michaelficarra]

Thanks @jmdyck. Done.