fix: fix f64 has no sufficient precision during parsing

[WenyXu]

I hereby agree to the terms of the GreptimeDB CLA.

Refer to a related PR or issue link (optional)

fix #3479

What's changed and what's your intention?

Enable the float_roundtrip of serde_json, it makes f64 -> JSON -> f64 produces output identical to the input.
However, this comes at an approximately 2x performance cost for parsing floats compared to the default best-effort precision.

See also:

• https://github.com/serde-rs/json/blob/master/
• Perfect accuracy float parsing serde-rs/json#536

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR does not require documentation updates.

[codecov]

Codecov Report

Attention: Patch coverage is 97.87234% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 85.03%. Comparing base (a309cd0) to head (ee4564b).
Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3483      +/-   ##
==========================================
- Coverage   85.45%   85.03%   -0.43%     
==========================================
  Files         896      897       +1     
  Lines      147532   148153     +621     
==========================================
- Hits       126080   125982      -98     
- Misses      21452    22171     +719     

[v0y4g3r]

LCTM

[tisonkun]

@WenyXu where do we rely on f64 to JSON to f64? If it's not on the hot path, the performance issue can be insignificant.

[WenyXu]

@WenyXu where do we rely on f64 to JSON to f64? If it's not on the hot path, the performance issue can be insignificant.

The column's default constraint will be serialized into bytes(JSON) when creating the table. However, these changes may also affect the insertion performance.