fix: only read the data file's fields from the page table and not the whole dataset's fields

[westonpace]

During the refactor to split things into lance-file and lance-table I introduced a bug. There were a number of code paths for opening a file reader and I consolidated them into one. However, when we were reading one data file from several data files, we ended up trying to load too many fields from the page table (more fields than exist in the page table).

Most of the time, this was ok. The read would read past the end of the page table and grab part of the schema (effectively garbage data). We never actually accessed this data and so we would get away with it. However, if the fragment schema was considerably larger than the data file schema, then the read would go past the end of the file and cause an error.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 88.78505% with 12 lines in your changes are missing coverage. Please review.

Project coverage is 80.68%. Comparing base (faed71b) to head (bbd219b).

Files	Patch %	Lines
rust/lance-file/src/reader.rs	90.32%	1 Missing and 8 partials ⚠️
rust/lance/src/index/vector/hnsw.rs	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2095      +/-   ##
==========================================
- Coverage   80.69%   80.68%   -0.01%     
==========================================
  Files         160      160              
  Lines       47022    47112      +90     
  Branches    47022    47112      +90     
==========================================
+ Hits        37946    38014      +68     
- Misses       6922     6928       +6     
- Partials     2154     2170      +16     

Flag	Coverage Δ
unittests	80.68% <88.78%> (-0.01%)	⬇️

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.