fix: json schema serializes field metadata

[albertlockett]

fixes: #3378

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.49%. Comparing base (26eb471) to head (5406a2d).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3379      +/-   ##
==========================================
+ Coverage   78.45%   78.49%   +0.04%     
==========================================
  Files         250      250              
  Lines       90189    90243      +54     
  Branches    90189    90243      +54     
==========================================
+ Hits        70758    70838      +80     
+ Misses      16525    16503      -22     
+ Partials     2906     2902       -4     

Flag	Coverage Δ
unittests	78.49% <100.00%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[westonpace]

Some minor suggestions but seems ok. I don't really know what we use the JSON serialization for so I'm not sure I can consider all downstream implications.

[westonpace]

Not super important but would it be possible to do skip_serializing_if = "HashMap::is_empty" and then skip the option? I'm not sure if that is more or less clear.]

Hmm...maybe this won't work on deserialization.

[albertlockett]

I was thinking we'd keep this as Option::is_none as it maintains consistency with how we're currently handling this for schema metadata

[albertlockett]

Some minor suggestions but seems ok. I don't really know what we use the JSON serialization for so I'm not sure I can consider all downstream implications.

Thanks @westonpace -- fwiw at least once use case we have for Json schema serialization is that lancedb cloud uses it (specifically in the response to calling .schema() on a remote table.