fix: use retry loop when loading deletion files

[westonpace]

When loading deletion files we were directly accessing ObjectStore::inner which bypasses the retry loop. This makes loading deletion files somewhat brittle and was causing issues in scans (which can load a lot of deletion files at the same time they are loading lots of other ranges).

This PR changes the deletion file load to use the ObjectStore and adds convenience methods to make this easier in the future.

In the future we should also move this load to use a FileScheduler instead of ObjectStore so that these deletion file loads do not violate our max-iops-per-scan but that can be done later.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 70.11494% with 26 lines in your changes missing coverage. Please review.

Project coverage is 77.92%. Comparing base (78d90a9) to head (1fd84de).
Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
rust/lance-io/src/object_reader.rs	72.72%	10 Missing and 2 partials ⚠️
rust/lance-io/src/local.rs	57.14%	2 Missing and 4 partials ⚠️
rust/lance-io/src/object_store.rs	76.92%	0 Missing and 6 partials ⚠️
rust/lance-table/src/io/deletion.rs	33.33%	0 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2899      +/-   ##
==========================================
- Coverage   77.93%   77.92%   -0.02%     
==========================================
  Files         231      231              
  Lines       70613    70612       -1     
  Branches    70613    70612       -1     
==========================================
- Hits        55031    55022       -9     
+ Misses      12462    12457       -5     
- Partials     3120     3133      +13     

Flag	Coverage Δ
unittests	77.92% <70.11%> (-0.02%)	⬇️

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.