Preserve embedded NUL bytes in TEXT values

[sbking]

Summary

• pass the encoded UTF-8 byte length to sqlite3_bind_text so embedded NUL bytes are not treated as a C-string terminator
• read sqlite3_column_text with its sqlite3_column_bytes length and preserve embedded NULs without changing the existing Emscripten UTF-8 decoding behavior
• add independent regression coverage for the bind and read paths

Testing

• yarn test under Node 20 and Chrome: 2,993 passed, 0 failed
• with the original implementation, both new tests fail across the storage/build matrix while the existing 2,114 assertions pass
• node --check src/sqlite-api.js

References

• SQLite's sqlite3_bind_text contract specifies that a negative byte length stops at the first NUL, while an explicit non-negative length preserves embedded NULs.
• SQLite's column-value contract defines sqlite3_column_bytes as the full UTF-8 byte length and recommends calling sqlite3_column_text before sqlite3_column_bytes.
• SQLite explicitly supports embedded NULs in stored TEXT, while documenting caveats for some SQL functions and CLI rendering.

Checklist

• I grant to recipients of this Project distribution a perpetual,
  non-exclusive, royalty-free, irrevocable copyright license to reproduce, prepare
  derivative works of, publicly display, sublicense, and distribute this
  Contribution and such derivative works.
• I certify that I am legally entitled to grant this license, and that this
  Contribution contains no content requiring a license from any third party.

[rhashimoto]

Thanks for the PR! I have one requested change and one question inline.

[rhashimoto]

address can be 0 if the column is SQL NULL, in which case we should return null.

[rhashimoto]

I don't think I understand this part. Why wouldn't we use TextDecoder on the whole buffer here?