feat: split large reads into smaller parallel reads

[westonpace]

We can often end up with large pages. This is easier in 2.1 than it is in 2.0 but it can happen in 2.0 as well. When this happens we make one large request into object storage. However, these large requests are problematic. Partly because we get less performance (less parallelism) and partly because they are prone to failure, especially timeout, when the object store is under stress. As a result, it is better to split requests into smaller requests.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 92.80000% with 9 lines in your changes missing coverage. Please review.

Project coverage is 78.61%. Comparing base (b8b486c) to head (ce7a38b).

Files with missing lines	Patch %	Lines
rust/lance-io/src/scheduler.rs	93.85%	2 Missing and 5 partials ⚠️
rust/lance-io/src/object_store.rs	66.66%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #3801    +/-   ##
========================================
  Coverage   78.60%   78.61%            
========================================
  Files         274      274            
  Lines      102487   102601   +114     
  Branches   102487   102601   +114     
========================================
+ Hits        80564    80656    +92     
- Misses      18743    18759    +16     
- Partials     3180     3186     +6     

Flag	Coverage Δ
unittests	78.61% <92.80%> (+<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.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.