improve float to string cast by ~20%-40%

[psvri]

Which issue does this PR close?

Closes #.

Rationale for this change

We can use ryu crate to cast floats to string for better performance. Ryu is used in crates like csv and serde-json for fast float to string conversion.

What changes are included in this PR?

I am using ryu to cast floats to string. The perf numbers are as below

cast f32 to string 512  time:   [18.055 µs 18.458 µs 18.932 µs]
                        change: [-26.059% -24.296% -22.401%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 14 outliers among 100 measurements (14.00%)
  8 (8.00%) high mild
  6 (6.00%) high severe

cast f64 to string 512  time:   [20.884 µs 21.154 µs 21.470 µs]
                        change: [-42.782% -41.722% -40.605%] (p = 0.00 < 0.05)
                        Performance has improved.

Are there any user-facing changes?

No

[psvri]

Let me know if I need to create an issue for the PR.

[jhorstmann]

Nice! I thought I benchmarked lexical vs ryu some time ago and did not see an improvement, but I can reproduce a 12% to 22% improvement here. 🚀

[psvri]

I experimented with iota crate as well for ints to string casts, but I didnt see any performance improvement when compared to lexical.

[Jefffrey]

Nice 🚀