Pass batch size to JSON reader using environment variable

[shrshi]

Description

The JSON reader set the batch size to INT_MAX bytes since the motivation for implementing a batched JSON reader was to parse source files whose total size is larger than INT_MAX (#16138, #16162). However, we can use a much smaller batch size to evaluate the correctness of the reader and speed up tests significantly.
This PR focuses on reducing runtime of the batched reader test by setting the batch size to be used by the reader as an environment variable.
The runtime of JsonLargeReaderTest.MultiBatch in LARGE_STRINGS_TEST gtest drops from ~52s to ~3s.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[shrshi]

/ok to test

[davidwendt]

This looks good. I did not realize the environment variable is just for the test. Thanks for doing that. What is the execution time for the test now? Can you try it with compute-sanitizer too?

Also, just a nit, can you change plain size_t to std::size_t in places where this PR changed code?

[shrshi]

This looks good. I did not realize the environment variable is just for the test. Thanks for doing that. What is the execution time for the test now? Can you try it with compute-sanitizer too?

The execution time has dropped from 52s to 3s now. With compute-sanitizer, the runtime is 123s.

[bdice]

CMake approval.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[shrshi]

/ok to test

[shrshi]

/merge