feat: switch grisu2 float-to-string algorithm to hybrid of xjb & zmij algorithms

[JairusSW]

Fixes #3012.

Changes proposed in this pull request:

• Switch to xjb-as which is an improvement over zmij-as
• Comply to the ECMA262 Specification with the exception of a trailing .0

Note: I wrote xjb-as with claude assisting me for an initial port. I then optimized by hand. Everything is carefully checked and passes over a trillion fuzz cases compared directly against Number::parse() in V8. It also reaches 100% code coverage and tests every edge case both I and claude could think of.

It's good quality stuff though, and I'm quite confident in that.

Earlier main-vs-PR formatter measurements, before the exact-integer string path (10m iterations per cycle):

Randomish input bucket	main	PR	Speedup
f64 buffered, ns/value	42.2	20.9	2.02×
f32 buffered, ns/value	36.0	17.8	2.02×
f64 string-returning, ns/value	79.4	59.6	1.33×
f32 string-returning, ns/value	72.0	55.7	1.29×

Earlier generated code size, before the exact-integer string path (a main function calling dtoa/ftoa):

Program uses	main	PR	Δ
f64 only	4,120 B	6,357 B	+2,237 B
f32 only	4,118 B	5,192 B	+1,074 B
Both	4,342 B	9,431 B	+5,089 B

Most of this is the formatter which is now ECMA262 compliant, so, it's more complicated.
The actual tables are 910 bytes (before) vs 872 bytes (this pr)

• I've read the contributing guidelines
• I've added my name and email to the NOTICE file

P.S. Sorry about the diff. Most of it is generated WAT fixtures.

[PaperPrototype]

Just curious here, why the exception of a trailing .0?

[JairusSW]

Just curious here, why the exception of a trailing .0?

Max and Dan made that decision so that floats are easily identifiable when converting them to a string. For example, you know that the string "1.0" is a result of f64.toString() while the string "1" is likely a result of i32.toString(). Slight deviation from ECMA262-spec, but it's not awful

[MaxGraey]

Can you rename xjb.ts to dtoa.ts and in comments mentioned about implementation based on xjb and zmij? xjb filename is very exotic and doesn't explains exactly what this file does

[MaxGraey]

Also plz add extra tests which cover new implementation-specific edge cases (I anready found a bug) and uncomment existing test cases which failed with grisu2