Reduce flush overhead in streaming stdout copies

[ryanfowler]

Summary

• Remove per-chunk flush() calls from async stdout/pager streaming so large responses can flow through without forcing a flush on every 16 KiB write.
• Increase the internal copy buffer to 64 KiB to reduce read/write overhead on long streaming responses.
• Add regression tests that verify the streaming helpers still copy all bytes while flushing only once at the end.

Testing

• cargo fmt
• cargo clippy --locked --all-targets --all-features -- -D warnings
• cargo test --all-features